What is the difference between prompt engineering and context engineering?

Prompt engineering crafts the instruction: wording, structure, examples. Context engineering designs everything the model knows when it works: retrieved documents, memory, codified business knowledge, assembled per moment. The prompt asks; the context informs. Reliable business AI is mostly won at the context level, with good prompts embedded in it.

What is loop engineering compared to prompt engineering?

Prompt engineering optimises one instruction you give by hand; loop engineering designs the autonomous system that decides what to prompt, when, and whether the result is acceptable, running on a schedule toward a defined goal. The prompts do not disappear; they get embedded in a system that issues them without you.

Is prompt engineering dead in 2026?

No: it moved indoors. Every loop and every context pipeline executes prompts, and their quality still shapes output. What ended is prompting as the whole job: the leverage moved up to what the AI knows (context) and what system drives it (loops), with instruction-craft embedded in both.

What is context engineering in simple terms?

Managing the AI's desk: making sure the right documents, memory and business knowledge are in front of the model at the moment it reasons. RAG, agent memory and codified knowledge files are its main tools. Most wrong answers about your business are desk problems, not model problems.

Which should my business learn first?

In order: prompt basics for everyone (a day), context engineering for anything answering about your business (your first real project), loop engineering once one governed agent runs well (the compounding step). Skipping levels is the expensive path: loops amplify whatever quality the lower levels feed them.

Are these formal disciplines or just buzzwords?

The labels are recent and will churn; the structure is real and durable: instruction quality, knowledge quality, system quality, each capping the next. Treat the terms as handles on that structure, and evaluate any new coinage the same way: which level is it naming, and is that your bottleneck?

What is harness engineering and where does it fit?

A sibling term from the developer world: designing the environment one agent runs inside (its tools, permissions, checks and working setup), sitting between context engineering and loop engineering in the stack: the harness equips the worker, the loop schedules and drives it. Harness work maps to the tool calling and governance layers; the kitchen version is fitting out the station itself. Expect the term to appear in vendor material; place it on the stack and it stops being confusing.

Can one person in a small business cover all three levels?

Realistically, yes, and many do: a technically curious ops manager can write the brief-pack (level 1), own the document hygiene and grounding relationship (level 2), and operate a partner-built loop (level 3). What one person should not do is build level 3 alone without the lower levels' evidence behind them; the levels are also a seniority ladder for the automation itself, and promotion by results applies to systems exactly as to staff.

What does agentic AI mean in simple terms?

AI that works toward a goal rather than answering a single prompt: it plans steps, uses tools, remembers context, checks results and adjusts, escalating to humans at the edges of its authority. Cruise control follows a setting; a chauffeur handles a destination. Agentic AI is the chauffeur.

What is the difference between agentic AI and generative AI?

Generative AI produces content when asked: text, images, code. Agentic AI uses that generative capability as a reasoning engine inside a larger loop of planning, tool use and feedback to complete tasks. Every agentic system contains generative AI; most generative AI use is not agentic.

What is the difference between agentic AI and an AI agent?

Agentic AI is the pattern; an AI agent is one system built on it. In practice the terms travel together: an agent is the worker, agentic describes how it works, and multi-agent systems are teams of them.

Is ChatGPT agentic AI?

As a chat window, no: it responds to prompts. Given tools, memory and goals, which modern AI platforms increasingly support, the same model becomes the reasoning engine of an agentic system. Agentic is a property of the architecture around the model, not of the model itself.

What are examples of agentic AI in small businesses?

Returns processing within policy limits, invoice chasing matched to customer history, CRM records maintained from real emails, supplier research and comparison, support tickets resolved end to end with escalation paths. The common shape: one bounded goal, scoped tools, humans on the exceptions.

Should my business adopt agentic AI now or wait?

Adopt the sequence, not the headline: prove value with assistants and simple automation first, then deploy one bounded agent on a measured workflow, then expand on the plumbing you have paid for. Our pillar guide, Agentic AI vs Workflow Automation, runs this decision with real ROI maths, and the 90-Day Roadmap turns it into a calendar.

What is the difference between agentic AI and RPA (robotic process automation)?

RPA scripts a computer to click and type like a human would: fixed steps, brittle when screens change, no judgment. Agentic AI reasons toward goals and adapts. They are not rivals in practice: RPA often serves as hands for legacy systems without APIs, while the agentic layer does the deciding. The distinction to hold onto: RPA follows steps a human recorded; agents choose their own.

Who are the main platforms and frameworks for agentic AI?

The model providers, Anthropic, OpenAI and Google, supply the reasoning engines; frameworks like LangGraph, CrewAI and the OpenAI Agents SDK structure agents and teams; automation platforms such as n8n, Zapier and Make now ship visual agent builders; and MCP standardises tool connections across all of them. Our AI Automation Stack guide tours every layer, and our n8n vs Zapier vs Make comparison covers the platform tier. The strategic point: the architecture pattern is stable even as the logos rotate, so learn the pattern.

What is an agent loop in AI?

The repeating cycle that makes AI autonomous: assemble context, reason and choose one action, execute it, observe the result, check whether the goal is met or a boundary hit, and repeat. Every agentic system runs this rhythm; quality differences between agents are mostly quality differences between their loops.

What is the ReAct pattern?

Reason plus Act: the 2022 research insight that interleaving a model's reasoning with real actions, think, act, observe, think again, grounds each thought in evidence rather than letting the model hallucinate past reality. It is the ancestor of most modern agent loops and the default rhythm of single agents.

How many cycles does an agent task take?

Typically a handful to a few dozen: a worked example can run eight cycles in about forty seconds. Cycle counts balloon when context is poorly assembled or stopping conditions are missing, which is a design smell rather than a task property; budgets per task keep the rhythm honest.

What is the difference between an agent loop and loop engineering?

The agent loop is the inner rhythm: one task, cycled to completion. Loop engineering designs the outer system: a scheduled, goal-driven loop that finds work, dispatches inner loops, verifies, remembers and continues indefinitely. Inner loops existed all along; the outer discipline is what got named in 2026.

Why do agents get stuck repeating the same steps?

Circling: the loop lacks loop-detection, step budgets or a route-level plan, so locally sensible cycles repeat at machine speed. Standard fixes are budgets, detection, an upfront plan and replan triggers.

Does a better AI model mean a better loop?

It helps the reasoning step and nothing else. Context assembly, observation honesty, done-checks, budgets and logs are engineering, not model quality, and they are where most real-world agent differences live. Feed and check the loop before paying for a bigger brain.

What is a stopping condition in an AI loop?

An externally checkable rule that ends or pauses a loop: the goal verifiably met (done-check), resources spent (budgets), a threshold crossed (boundaries routing to humans), or a failure pattern detected (circuit breakers). The defining property: checkable outside the AI's own opinion of itself.

Why can't the AI just decide when it's finished?

Because a model's done is a fluent claim, produced by the same mechanism as its correct and incorrect answers alike. Trustworthy done-checks use mechanical verification or a separate checker model, the maker-checker split applied to completion itself, which is exactly how the leading coding-agent tools implement their run-until-done features.

How do I prevent runaway AI costs in loops?

Four layers: hard budgets per task and per run (cycles, tokens, actions, time), cost-velocity alerts (spend rate versus baseline), context discipline so cycles do not bloat, and discovery sanity bounds so a data hiccup cannot feed the loop an ocean. Budgets are the layer that works when everything else surprises you.

What is a circuit breaker in AI automation?

A stop triggered by failure patterns rather than single events: repeated check failures, recurring tool errors, quality collapse or abnormal spend velocity, pausing the loop and alerting a named owner. Borrowed from electrical engineering for the same reason: patterns of fault deserve automatic disconnection.

What should be in a loop's stop-sheet?

One page: the verifiable done-condition per item and per run, the four budgets (cycles, tokens, actions, time), the boundary thresholds routing items to human approval, the circuit-breaker rules, and the named owner of halts with a one-line runbook. If it does not fit a page, the loop's scope is probably too broad.

How do stopping conditions relate to human approval gates?

Gates are one of the four stop types (boundaries): conditions that route an item to a person rather than ending the run. The gate-map design from our human-approval guide slots directly into the stop-sheet; loops simply give the gates a schedule and make their absence more expensive.

What is maker-checker in AI loops?

The structural rule that the agent producing work is never the one verifying it: a second agent, differently instructed and ideally a different model, reviews every item against explicit criteria before the loop proceeds, and completion itself is judged by something other than the worker. In unattended loops this split is not best practice; it is the component that replaces the absent human.

How do you verify AI output when nobody is watching?

Three built layers: mechanical checks on every testable fact at action time, an independent checker agent per work item, and verified done-conditions per run, all writing to a catch log, backed by a permanent human sampling habit that measures the escape rate. Designed before walking away, or it does not exist.

The record of everything verification intercepted: mechanical failures, checker rejections, done-check disputes, with reasons. It proves the checks are alive, supplies the tuning data since catches cluster around real causes, and answers audits. An active loop with an empty catch log has dead checks, not a perfect agent.

Can two AIs checking each other really be trusted?

More than one AI trusted alone, less than the marketing implies: different model families reduce shared blind spots, mechanical verification outranks both, and sampled human review remains the epistemic floor. The design goal is not certainty; it is a known, small, monitored escape rate with bounded consequences.

How much does loop verification cost?

Mechanical checks are effectively free; a checker pass adds a modest model call per item, pennies; the human sample costs minutes on a cadence. Against unattended mistakes at machine speed, it is the most lopsided trade in the discipline, the reflection economics compounded by schedule.

What happens when the checker and maker disagree repeatedly?

Revision caps end the ping-pong, two redrafts then the human queue, and persistent disagreement is treated as signal, not friction: work the loop cannot satisfy itself about is precisely what a person should see. A spike in disagreements trips the quality breaker and pauses the run, because it usually means the world changed upstream.

Who in a small business should own loop verification?

The same person who owns the loop: usually whoever held the responsibility before it was automated. Their verification duties are concrete and bounded: the daily exception queue (minutes), the weekly catch-log read (fifteen minutes), the monthly sample, and the quarterly checklist review. It is a management role over a tireless report, not a technical one, and the businesses that thrive at this treat it as a named part of a real job description rather than a hope.

Do AI agents have memory built in?

No. The underlying models are stateless: they forget everything between requests. All agent memory is engineered around the model in layers: a managed context window (working memory), databases for cross-session facts and history (long-term), and retrieval from your documents (knowledge). Quality varies exactly as much as that engineering does.

What is a context window in simple terms?

The model's notepad: everything it can consider at once, from instructions to conversation history to fresh data. It is large but finite, and managing what deserves space on it, verbatim, summarised or archived, is a core part of agent design.

What is the difference between RAG and agent memory?

RAG is one layer of it: retrieving relevant knowledge from your documents at question time. Full agent memory adds working-memory management and long-term episodic and factual storage about specific customers and tasks. RAG answers what is true about the business; long-term memory answers what happened with this customer.

Why does my AI keep forgetting things mid-conversation?

Notepad overflow: the conversation outgrew the context window and something got summarised away or dropped. Better builds summarise selectively and push details to retrievable storage rather than losing them. If it forgets across sessions, the issue is absent long-term memory instead.

Is agent memory a privacy risk?

It is where the privacy work lives: memory persists customer data, so retention limits, access scoping, audit logging and deletion policies all attach here. Well-designed memory is auditable and forgettable by policy; undesigned memory is a data-protection finding waiting for its auditor.

Can an AI agent learn from its mistakes?

Yes, through memory plus reflection: outcomes get recorded, corrections get written back as curated facts, and future retrievals supply the lesson. It is bookkeeping rather than mystical learning, which is good news: bookkeeping is reliable, inspectable and yours.

How much does agent memory infrastructure cost?

Modestly, by AI standards: long-term memory runs on ordinary databases, often PostgreSQL already in your stack, and retrieval infrastructure, embeddings plus a vector store or pgvector inside the same database, adds only modest monthly hosting at SMB scale, not enterprise line items. The real cost is design and curation time: deciding what gets written, kept and forgotten. Design is a one-off and curation is a habit, while infrastructure bills recur.

Should different agents share memory?

Share the library, since business knowledge should be consistent everywhere, and be deliberate about the filing system: shared customer records need ownership rules for writes, while task-specific episodic memory usually stays per-agent. The failure mode to avoid is every agent writing freely to shared memory, where one agent's mistaken conclusion becomes every agent's confident context.

What Is an AI Agent? The Anatomy, Piece by Piece

Aug 20

Written By Alpa Parmar

Open the bonnet on an AI agent: the reasoning engine, tools, memory, feedback loop and boundaries that make one work, how a single task flows through them, and what to check before you buy one.

You can use a car for years without knowing what a crankshaft does. But the day you buy one, a little mechanical literacy changes everything: you stop nodding at the salesperson and start asking the questions that reveal what is actually under the paint.

AI agents have reached that moment for business buyers. The demos are shiny, the word "agent" is on everything, and the difference between a solid one and a flashy liability is entirely internal. So this article opens the bonnet: the five components every real agent has, how a single task actually flows through them, where each component fails, and the checks that tell you whether a vendor's agent has all its parts.

(If you want the concept before the mechanics, start with [What Is Agentic AI?]; if you want the buying comparison against chatbots, that is [AI Agents vs Chatbots]. This article is the engineering tour, in plain English.)

The Five Components of Every Real Agent

1. The reasoning engine: the brain. A large language model (Claude, GPT or a peer) that interprets the goal, breaks it into steps, and decides what to do next at every junction. This is the component you rent rather than own, paying per use, and the one everything else is built around. Worth knowing: good agent builds often use different models for different moments, a cheap fast one for routine steps and a stronger one for the judgment calls, the way a firm staffs juniors and seniors.

2. Tools: the hands. Connections to real systems: the CRM lookup, the email send, the calendar write, the payment API. Each tool is a specific capability with specific permissions, increasingly plugged in through the open MCP standard (the USB-C of AI, explained in our [MCP guide]). An agent's real-world usefulness is roughly the quality of its toolset; its real-world risk is roughly the breadth of it. [Tool Calling Explained] covers the mechanics.

3. Memory: the notebook. What the agent knows beyond the current moment: the conversation so far, the customer's history, the business rules, what it already tried on this task. Memory is layered (working, long-term, and knowledge retrieved from your documents), and its design quietly determines whether the agent feels competent or amnesiac. The full tour is in [Agent Memory Explained].

4. The feedback loop: the eyes. After every action, the agent observes the result: did the API call succeed, does the data look sane, did the step achieve what it was for? Then it adjusts: retry differently, flag the oddity, choose another path. Upgraded with deliberate self-review, this becomes [reflection], the single biggest reliability difference between production agents and conference demos.

5. Boundaries: the job description. Scope ("returns under $200"), refusals ("never discuss legal matters"), permissions per tool, spending limits, and the escalation path to a human. Boundaries are not a restraint bolted onto the agent; they are a component of it, as much as the brain. An agent without them is not more powerful; it is unfinished. ([AI Agent Governance] is the full manual; [Human Approval] covers the gates.)

The analogy that holds it together: an agent is a capable new hire, anatomically. Brain, hands, notebook, eyes, and a signed job description. Interview a human candidate and you assess all five; evaluate an agent and you should too, because the failure of any one produces a recognisable workplace disaster: clever but handless (all talk), handy but blind (accidents), sharp but amnesiac (repeats itself), and brilliant but boundaryless (the intern with the master key).

How a Task Actually Flows: One Return Request, Start to Finish

Anatomy is clearer in motion, so let's follow one task through all five components. A customer emails: "This jacket doesn't fit, I'd like to return it."

Step 1: The goal meets the brain. The reasoning engine reads the email against its standing goal ("process return requests within policy") and plans: identify the order, check eligibility, decide the route, act, confirm.

Step 2: The notebook opens. Memory supplies context: this customer's order history, the returns policy (retrieved from your documents via the [RAG pattern]), and a relevant note: this is their second return this quarter.

Step 3: The hands move, one tool at a time. Tool call one: order lookup (read-only permission). Tool call two: policy check against the order date. Each call is logged: which tool, what request, what came back.

Step 4: The eyes check. The order lookup returned two candidate orders (the customer has two recent jackets). The feedback loop catches the ambiguity: rather than guessing, the agent drafts a one-line clarifying question to the customer. A lesser build would have picked one and been wrong half the time, politely.

Step 5: The boundaries speak. Customer clarifies; the return is eligible; the refund value is $240. That crosses the $200 approval threshold, so the agent does not act: it packages the case (order, policy clause, history, recommendation) and routes it to a named human for one-click approval. ([Human Approval] explains why this gate design is the whole art.)

Step 6: Action, confirmation, record. Approved. The agent generates the label (write permission, scoped), emails the customer, updates the system, and logs the complete trail. Total human involvement: one click and one glance. Total customer experience: fast, accurate, and indistinguishable from a very good employee.

Six steps, five components, each doing exactly its job. That is what "an AI agent" means, mechanically.

Why Should a Business Buyer Know the Anatomy?

Because demos show the brain and hide the rest. Every demo showcases reasoning: the impressive plan, the fluent reply. What separates production agents from demos is everything else: the memory design, the feedback handling, the boundary work. Anatomy literacy redirects your attention to where quality actually lives.

Because the anatomy is the quote. When a vendor prices an agent build, you are paying for components: tool integrations (usually the biggest line, per our [pricing guide]), memory infrastructure, and governance plumbing. Understanding the parts is understanding the invoice.

Because failure diagnosis follows the anatomy. When an agent misbehaves, the fix depends on which organ failed: wrong facts is a memory problem, wrong action is a boundary problem, gave-up-too-easily is a feedback-loop problem, nonsense plan is a reasoning problem (often just a model-tier problem). Teams that know the anatomy fix in days what mystery-box owners abandon in months.

The Diagnostic Table: Symptom to Organ

Anatomy earns its keep the day something misbehaves, so here is the troubleshooting map, symptom by symptom.

"It stated something untrue about our business." Memory, usually the library layer: a stale or missing document, or retrieval fetching the wrong page. Fix the document or the retrieval, not the model. If it invented something no document covers, the missing "say I don't know" instruction is the culprit.

"It forgot what we discussed earlier." Memory, working layer: notepad overflow with careless summarisation. Across sessions, absent long-term memory. Both are build decisions, both fixable.

"It did the steps in a strange order, or went in circles." The brain's planning, missing route discipline: step budgets, loop detection, an upfront plan. [Agent Planning] covers the fixes.

"It gave up the moment something failed." Feedback loop too timid: no retry logic, no alternative paths. The opposite symptom ("it kept hammering a broken API for an hour") is the same organ with no budget.

"It did something it should never do." Boundaries: either the scope was never written, or the toolbox was overstocked (a tool it should not hold), or a gate was missing. This one is a governance incident, not a bug: fix the boundary, then ask why it was absent.

"It is right but slow and expensive." Usually the brain doing all steps on the premium model, or wandering plans burning tokens. Model routing and planning discipline are the levers.

"It works but nobody trusts it." Not an organ at all: missing evidence. Baselines, logs and catch records ([reflection]'s output) are what convert a working agent into a trusted one.

Tape this table next to any agent deployment and most "the AI is broken" conversations become fifteen-minute fixes with the right person. Mystery is optional.

What Are the Limitations of Even a Complete Agent?

Competence is narrow. All five components are built around a job. The superb returns agent knows nothing about your invoices; agents generalise across tasks about as well as job descriptions do. Bigger goals need teams of specialists, which is the [Multi-Agent Systems] chapter.

The brain is rented and probabilistic. Reasoning engines occasionally reason wrongly, with confidence. The surrounding anatomy (grounded memory, feedback checks, boundaries, human gates) exists precisely because the brain alone cannot be fully trusted, and never will be. Design assumes fallibility; that is the design working, not failing.

Every component ages. Models update, APIs change, policies evolve, memory fills with the stale. An agent is a system to maintain, not a purchase to forget: the month-thirteen question applies to every organ.

Cost lives in the loop. Each reasoning step burns tokens; long feedback loops on hard tasks burn many. Well-built agents route easy steps to cheap models and reserve the expensive thinking, which is invisible in demos and very visible in year-one running costs.

The Buyer's Anatomy Check

Six questions, one per component plus one for the whole:

Brain: which model(s), and can the routine steps run on cheaper ones?
Hands: exactly which tools, with which permissions, over which standard (MCP or bespoke)?
Notebook: what does it remember across sessions, where does that live, and is the knowledge grounded in OUR documents?
Eyes: show me what happens when a tool call fails or returns nonsense.
Job description: show me the written scope, the gated actions, and yesterday's audit log.
The whole: who maintains each part, and what does month thirteen look like?

A vendor fluent in all six is selling an agent. Fluent in one is selling a demo with a brain.

Frequently Asked Questions

What is an AI agent in simple terms? Software that completes tasks toward a goal: a reasoning engine (brain) plans the steps, tools (hands) act on real systems, memory (notebook) holds context, a feedback loop (eyes) checks results, and boundaries (job description) keep it inside its authority with humans on the exceptions.

How is an AI agent different from ChatGPT? ChatGPT in a chat window is a brain alone: brilliant conversation, no hands, no persistent job. An agent wraps that same class of brain with tools, memory, feedback and boundaries so it can actually complete work. The model is one component of five, not the product.

What are the main components of an AI agent? Five: the reasoning engine (an LLM), tool connections to real systems, layered memory, a feedback loop that observes and adjusts, and explicit boundaries with escalation paths. Production quality lives mostly in the last four, which demos rarely show.

Do AI agents make mistakes? Yes, and well-built ones are designed around that fact: grounded memory reduces wrong facts, feedback loops catch failed steps, boundaries cap the damage of wrong actions, and human gates hold the consequential calls. The realistic promise is not zero errors; it is small, caught, reversible, explainable ones.

How much does an AI agent cost to build? For SMB single-workflow agents: typically £8,000 to £25,000 to build plus running and maintenance, with tool integrations usually the largest line (full tiers in [How Much Does AI Automation Cost?]). The anatomy explains the spread: more tools, richer memory and stricter governance all add parts.

Can I build an AI agent without coding? Increasingly yes, for moderate cases: platforms like n8n ship visual agent components (brain node, tool nodes, memory), covered in our [n8n vs Zapier vs Make] comparison. The anatomy stays identical; only the assembly method changes. Complex, regulated or high-volume agents still justify custom builds, per the pillar's decision rules.

How long does it take to build an AI agent? For a bounded single-workflow agent with the full anatomy: typically four to eight weeks, of which the brain is days and everything else is the schedule: tool integrations (longest, and dependent on your systems' APIs), memory design, boundary definition and the shadow-testing period that earns launch. The [90-Day Roadmap] wraps this in the audit and measurement that make the build worth doing. Timelines quoted in days are quoting a demo; timelines quoted in quarters, for one workflow, deserve a second opinion.

Do I need a different agent for every task? Broadly, one agent per coherent job description, which usually means per workflow: the returns agent, the invoice-chasing agent, the intake agent. They share plumbing (connections, memory infrastructure, governance patterns), which is why the second agent costs roughly half the first. Stretching one agent across unrelated jobs produces the fuzzy scope that the [failure statistics] punish; hiring a team of specialists is the [multi-agent] chapter.

The Takeaway

An AI agent is not magic and not a model: it is five parts working as one worker. Brain to plan, hands to act, notebook to remember, eyes to check, and a job description that says where it stops and a human begins.

Learn the anatomy once and everything downstream gets easier: demos stop dazzling you, quotes start making sense, failures become diagnosable, and the six buyer's questions do the rest. The bonnet is open for the rest of the series: [memory], [planning], [reflection], [tool calling] and [the human gates], one organ at a time.

Bots and Brand Works builds agents with all five parts and shows you each one working before launch: the scope, the permissions, the logs, the gates. Considering a build or holding a quote? Send it over and we will tell you which components it prices and which it forgot.

Need Help Implementing AI?

Resources and Further Reading

Pillar article: Agentic AI vs Workflow Automation: The 2026 Enterprise Guide
Related: Tool Calling Explained [add internal link]
Related: What Is Agentic AI?
Also in this series: Agent Memory · Agent Planning · Reflection · Human Approval [add internal links]
Anthropic: Building effective agents: Click Here
Model Context Protocol: Click Here

Alpa Parmar

AI Marketing Automation

https://www.botsandbrandworks.com