Skip to the intent console
Akkordintent · agents · integration

This site has
no menu.

Say what you need built and it assembles the answer: the position, the surfaces, the run, the stack. That is also the thing we build for other people: interfaces that resolve intent instead of filing it under a heading.

intent router · idle

what actually differs

Most studios ship you a unit of work. We ship a motion that keeps improving.

Website people optimise for landings and traffic: how many arrived. We optimise the loop that starts after they arrive: getting the attention of someone who can actually buy, then removing every step between interested and paying. We watch where people drop off, change the surface, and measure whether the change worked. That runs continuously, not once at launch.

the secret weapon

More why questions than anyone asks for. Most briefs describe a solution someone already picked; the job is to get back to the intent underneath it. After that, every interaction and micro-action on the thing we build becomes data about whether we got it right.

the loop, left to right/keep scrolling

  1. 01Attention

    Not more traffic. The attention of someone who can actually buy. Volume is the easy metric and usually the wrong one.

    Qualified arrivals, not sessions

  2. 02Why, repeatedly

    We ask why until the real need surfaces. A brief almost always describes a solution someone already chose; the intent underneath it is the thing worth building for.

    Stated brief vs. actual intent

  3. 03Friction

    Every step between interested and paying gets counted, then cut. Complexity is where intent goes to die, and it is nearly always removable.

    Steps to value, time to value

  4. 04Signal

    Every interaction and micro-action leaves data. Drop-off becomes something you can see and point at, rather than something you infer after a bad quarter.

    Drop-off per stage

  5. 05Motion

    The surface adapts on what the data says, and we measure whether the change earned its place. Launch is where this starts, not where it stops.

    Change vs. movement in the loop

The loop closes, then starts again.

before anything gets built

The brief is never the objective.

Every one of these arrived describing a solution. None of them was wrong to ask; each was just answering a question nobody had checked yet. The distance between the left column and the right is the whole job.

  1. 01as it arrived
    “We need a chatbot to answer customer questions.”
    router reads this asAgent systems

    what we asked

    • Who is asking, and what do they actually ask?
    • What happens after they get an answer?

    what that surfaced

    Nearly every question was the same handful of pre-qualification questions. The cost was not answering them; it was that anything arriving after five in the afternoon sat until the next morning, and by then a good share of those people had booked elsewhere.

    the actual objective

    Cut the gap between a question and a next step to under a minute, at any hour, without inventing an answer.

    what got built

    An agent that reads the real availability and pricing systems, answers from them, and books the next step in the same exchange. It says it does not know rather than guessing, and hands to a human on anything it cannot source.

    first move

    Two weeks tracing one week of real inbound questions end to end, to find which ones actually stall a decision.

  2. 02as it arrived
    “Can you make our website rank better and bring in more traffic?”
    router reads this asIntent-based web

    what we asked

    • What happens to the traffic you already have?
    • Of the people who enquired last month, how many could actually buy?

    what that surfaced

    Traffic was never the problem. The enquiry form asked for eleven fields including a budget range, and most people who started it left at the budget question. More traffic would have bought more of the same drop-off.

    the actual objective

    Raise qualified enquiries from the traffic already arriving, and know which step loses people.

    what got built

    The path from arrival to enquiry rebuilt around what someone can answer immediately, with the budget question moved to the conversation where it belongs, and drop-off measured per step rather than per page.

    first move

    Two weeks instrumenting the funnel that exists, so the first change is aimed at a measured step rather than a guessed one.

  3. 03as it arrived
    “We keep hearing we should have an AI model in the business somewhere.”
    router reads this asAI development

    what we asked

    • Which decision is currently slow, or wrong often enough to hurt?
    • Where does the knowledge behind that decision actually live?

    what that surfaced

    Quoting took three days, because the pricing logic lived in one person's head and everyone waited for them. Not a technology gap: a single point of failure everyone had learned to work around.

    the actual objective

    Get a defensible quote out in under an hour without that person in the loop, and without inventing numbers.

    what got built

    Retrieval over several years of quotes and their outcomes, drafting a priced quote with its reasoning shown and every figure traced to a comparable job, plus a review step before anything reaches a client.

    first move

    Two weeks on one quote type, measured against what the expert would have quoted, written up whether or not it works.

  4. 04as it arrived
    “Our systems do not talk to each other and we need them integrated.”
    router reads this asService integration

    what we asked

    • What does somebody retype, and how often?
    • What happens today when it goes wrong?

    what that surfaced

    Bookings were retyped into the accounting system twice a day. The retyping was survivable; the invisibility was not. Failures surfaced weeks later during reconciliation, by which point nobody could say which record was right.

    the actual objective

    Remove the retyping, and make a failure visible the hour it happens instead of at month end.

    what got built

    A one-way sync with reconciliation, and a failure queue somebody owns: anything that does not land shows up as a row a human can retry or reject, rather than a silent gap.

    first move

    Two weeks on the single highest-volume record type, including the failure path, because that is the part that decides whether this gets trusted.

  5. 05as it arrived
    “We just need something simple to start with, nothing fancy. What would it cost?”
    router reads this asEngagement

    what we asked

    • What has to be true in six months for this to have been worth doing?
    • Who do you need to believe you, and what would convince them?

    what that surfaced

    They needed two enterprise clients to sign. A brochure site would not have moved that at all, and would then have been judged a failure for not doing a job it was never built to do.

    the actual objective

    Give a sceptical enterprise buyer enough evidence to take a first meeting.

    what got built

    Not a brochure. The proof those buyers ask for, put where they look for it, with the enquiry path built for someone doing due diligence rather than someone browsing.

    first move

    Two weeks tracing one real sales conversation end to end, before anything is designed. If it turns out a quick site was right, that is a cheap way to find out.

These are patterns, not case studies. Every brief on the left arrives in this shape regularly; none of them is a claim about a particular client.

Four practices. One spine.

Every row below is an intent. Press one and the console resolves it.

Currently building for

3 active engagements

An answer, taken apart.

Everything above the model is composed, not generated. The order is the point.

  1. 01Your briefYouPlain language, up to 600 characters. No form, no category, no dropdown.
  2. 02Signal matchDeterministicStem matching against the four practices. Zero network calls, single digit milliseconds.
  3. 03The written positionDeterministicWhat the studio actually thinks about this problem, written once by a person.
  4. 04Model rewriteOptionalRewrites the prose to your wording. It never touches the surfaces, the run or the stack.
  5. 05The surfaceComposedThree voices: position, capability surfaces, and a proposed run with time estimates.

in progress

Property is the loop at its worst.

It is the highest-intent purchase most people ever make, and the journey from first interest to signed transfer is measured in weeks of dead air, duplicate admin and portals that answer a question nobody asked. That is the exact shape of problem the studio is built for, so proptech is where we are spending our own time.

  • Search by intent, not by filter

    Nobody wants three bedrooms. They want a school run under fifteen minutes and a room that fits a desk. Filters make the buyer translate their life into a database schema. The translation should run the other way.

  • Qualification that respects everyone

    An agent that answers the boring questions immediately, at any hour, and books the viewing only when the match is real. Fewer wasted Saturdays for the buyer, fewer dead leads for the agency.

  • The spine between portals, CRM and conveyancing

    The listing, the enquiry, the offer and the transfer file live in four systems that disagree. Most of the delay in a property transaction is not legal work; it is waiting for information to move.

  • Drop-off measured across weeks, not sessions

    A property funnel spans months and several parties. Instrumenting it end to end is harder than a checkout and considerably more valuable, because every stalled week is real money.

None of this is a launched product. It is where our own build time is going, and the reason we are interested in talking to people who work in property.

machine surface

Some of your visitors are not people.

So this site answers both. The same router that renders the interface above is exposed as a streaming endpoint and described in /llms.txt. An agent doing research on someone’s behalf gets structured events, not a scrape of our marketing copy. And if it arrives inside a WebMCP-capable browser, it does not have to read the page at all. This document hands it tools.

POST /api/intent
curl -N https://akkord.tech/api/intent \
  -H 'content-type: application/json' \
  -d '{"intent":"an agent that triages inbound email into our CRM"}'

application/x-ndjson

  • {"type":"resolved","kind":"agents","confidence":0.81}
  • {"type":"delta","text":"A chat window is not an agent. "}
  • {"type":"surface","surface":{"id":"a1","label":"Durable execution", … }}
  • {"type":"step","step":{"actor":"architect","action":"Draw the boundary", … }}
  • {"type":"stack","stack":["durable workflows","MCP tool servers", … ]}
  • {"type":"done"}
machine trafficsampled while quiethumans counted, never listed

non-human interactions

0.0per minute

  • 0agents
  • 0crawlers
  • 0other bots

last 2 min

peak 1

  • Listening.
WebMCPchecking8 tools

Looking for a model context on this document.Registered via document.modelContext.registerTool. The tools read from the same source the page renders, so an agent and a person get the same answer, and akkord_submit_intent_to_page lets an agent drive the console above on your behalf.

  • akkord_resolve_intent
  • akkord_list_practices
  • akkord_list_clients
  • akkord_get_engagement_model
  • akkord_how_we_work
  • akkord_contact_channels
  • akkord_open_contact_channel
  • akkord_submit_intent_to_page

Straight answers.

What does Akkord do?
Akkord is a software studio building four things: intent-based websites, AI agent systems, applied AI development such as retrieval and extraction, and integrations between services. The common thread is reducing the distance between what someone wants and getting it.
How is Akkord different from a website agency?
Website agencies optimise for landings and traffic, meaning how many people arrived. Akkord optimises the loop that starts after they arrive: cutting the steps between interested and paying, measuring drop-off at each stage, and changing the surface based on what the data says.
How does an engagement with Akkord start?
With a two-week diagnostic. We trace your path from first attention to paying customer end to end and measure the drop-off at each step rather than guessing it. You keep the write-up whether or not you continue.
Does Akkord build AI agents?
Yes. Agents with durable execution, scoped tools over your real services, a permission boundary agreed before any code is written, and a readable trace of every decision. They escalate to a person when a call is above their pay grade.
Can an AI agent use the Akkord website directly?
Yes. The site exposes a streaming intent endpoint at /api/intent, a description at /llms.txt, and WebMCP tools registered on document.modelContext, so an agent can resolve a brief or open a contact channel without scraping the page.
Who does Akkord work with?
Current engagements include Outercite in AI search intelligence, getohdit in legal and digital risk for South African law firms, and ALM Digital Media in travel and hospitality marketing across the UK, EU and South Africa.