The corpus

Everything we know about putting an agent in a consumer app.

59 posts, arranged as 8 regions rather than a feed. Most people arrive here from a search result on one deep post, so every post says where it sits and what to read next. If you arrived on purpose, start with the region that matches your question.

Start at the beginning The Agent basics guide

How this is organised

One region per question a team actually asks on the way to shipping. Every post belongs to exactly one, links the post that opens it, and ends by pointing at the next thing to do.

Posts: 59
Regions: 8
Since: 2024

The eight regions

Each opens with one long post and is filled in by the rest.

Agent basics
What is an in-app agent, and is it different from the chatbot we already tried?
Start here
What an in-app AI agent actually is, and what it can touch
10 posts
Render spec
How does an agent show a balance, a list or a form instead of describing it?
Start here
Generative UI on native mobile: how an agent draws real product UI
7 posts
Integration
What does putting this in my app actually involve?
Start here
How far you actually get in a day with an in-app agent
3 posts
Tools and MCP
How does the agent reach the backend I already run?
Start here
Connect an AI agent to the backend you already run
5 posts
Voice
What makes spoken turns feel fast enough to keep using?
Start here
Designing an agent conversation people can actually use
6 posts
Language
What breaks when the user is not speaking English?
Start here
How to evaluate dialect coverage in a speech stack
6 posts
Safety
How do I let an agent act without letting it act wrongly?
Start here
Confirm cards: letting an agent act without losing control
7 posts
Business
Is this worth building, and how would we know it worked?
Start here
How to add an AI agent to a mobile app: the complete guide
15 posts

Latest

24 of 59

12 Sept 2026An in-app agent is an accessibility feature, narrowlyAn agent that completes a task is an alternative input path, and one WCAG 2.2 criterion turns the biometric confirm into an accessibility win rather than a tax.Voice10 min
12 Sept 2026Does your agent work eight times out of eight?Average accuracy is the wrong number for a product. Run the same task eight times and count how often it worked every single time.Safety11 min
12 Sept 2026The first 300 milliseconds of an agent turnYou cannot make the model faster from the client. Almost all of the felt wait is decided before the model has said anything, and that part is yours.Render spec10 min
12 Sept 2026Generative UI on native mobile: how an agent draws real product UIGenerative UI on the web streams React components. On iOS and Android it has to become a typed spec over a fixed native catalogue. Here is why, and how it works.Render spec15 min
12 Sept 2026Prompt injection when the agent can spend moneyIn a consumer app the untrusted text is your own user's data and the tools move their money, which makes filtering useless and structure the only real defence.Safety10 min
12 Sept 2026Server-driven UI was already the answer. LLMs made it urgent.Server-driven UI and generative UI are one idea arriving from two directions, and a decade of mobile practice already tells you which parts are hard.Render spec10 min
12 Sept 2026What an in-app AI agent actually is, and what it can touchThree different things get called an AI agent in a mobile app. Here is the one that lives inside your product and acts through your own backend.Agent basics11 min
12 Sept 2026When not to put an AI agent in your appFour situations where an in-app agent loses to the interface you already have, and the test to run before you commit a quarter to building one.Agent basics10 min
12 Sept 2026Where the seconds go in an in-app agent turnWe measured a slow turn end to end. Most of the time was not in our code and not in the model, which changes what is worth optimising.Tools and MCP10 min
12 Sept 2026Which agent actions deserve Face ID, and which do notBiometric-gating everything trains users to approve without reading. Tier agent actions by what they can destroy, and let the platform decide how.Safety10 min
12 Sept 2026A catalogue beats free-form generation, and the objections say whyEvery serious objection to generative UI is an objection to free-form generation. A fixed catalogue answers all four without giving up the idea.Render spec10 min
28 Jul 2026Letting users reorder by voice in three tool callsRepeat purchase is the cheapest first agent feature to ship. Three tools carry it, only one of them writes, and a confirm card sits between the second and the third.Agent basics6 min
24 Jul 2026How a render spec turns one JSON payload into native UIThe whole contract is a spoken answer, a separator, and an array of widget objects. The interesting parts are the constraints, not the format.Render spec10 min
21 Jul 2026Agent or chatbot: telling the two apart before you buyA support bot answers questions. An in-app agent finishes the task. Five questions that separate them in any vendor demo, and where a chatbot still wins.Agent basics10 min
17 Jul 2026Running ten dialects of one language through one pipelineMost speech stacks assume one user speaks one language. Production breaks that in week one, and the fixes are the same whichever language you start from.Language10 min
14 Jul 2026Confirm cards: letting an agent act without losing controlIf the confirmation lives in the prompt it is a suggestion. Make it a branch in your server code, and decide the tier per tool rather than per sentence.Safety12 min
10 Jul 2026Connect an AI agent to the backend you already runYou do not need a new API surface for an agent. You need to describe the one you have, and decide what it is never allowed to touch.Tools and MCP11 min
10 Jul 2026MCP for in-app agents: connecting your tools to a modelThe Model Context Protocol replaces one connector per backend with one protocol. What it is, how a tool call actually happens, and what it costs on the first turn.Tools and MCP10 min
07 Jul 2026What barge-in really costs, and how we pay for itServer-side turn detection decides when an agent should stop talking. The client decides whether it actually does, and that is where it breaks.Voice11 min
03 Jul 2026Designing a tool surface an agent can actually useTwo independent vendors put the practical ceiling near twenty tools. Everything else about tool design follows from the model reading your schema and nothing else.Tools and MCP9 min
30 Jun 2026What the first week of real assistant traffic teaches youReal users ask for things no test plan contains. Three questions only live traffic answers, and the metric that looks good while telling you nothing.Business10 min
13 Jun 2026How to add an AI agent to a mobile app: the complete guideWhat an in-app agent is made of, what each of its five pieces costs to own, and the honest path from a first spoken turn to something you can hand to users.Business12 min
13 Jun 2026Where in-app agents are being adopted fastestAdoption is running ahead of the English-speaking world in several large markets. What that means for a team shipping an agent outside its home country.Business10 min
13 Jun 2026In-app search when the query is a whole sentenceHow search inside an app breaks when people speak in sentences, and how to build the search tool an agent can actually call.Render spec7 min

By kind

Guides22
Engineering25
Product6
Market Insights6

By topic