For independent performance, growth & creative agencies (15–100 people)

Senior teams were hired to think.
Sundays are for formatting decks.

We rebuild the workflows where agencies leak margin — pitch decks, QBRs, brief intake, performance commentary, approval flows. Audit-first. Working prototype on your real data in weeks. Not a retainer, not a workshop deck.

Bring one workflow. Leave with a yes/no on whether it can pay back in six weeks.

One agency week · before / after Manual cycle With maxodev
Tue
Pitch deck assembly
Before
8 h
After
45 min
7h 15m back to senior
Wed
Brief intake & scoping
Before
3 d
After
90 min
22h 30m back to senior
Thu
Client QBR draft
Before
6 h
After
by 8am
5h 15m back to senior
Fri
Performance commentary
Before
16 h
After
3 h
13 h back to senior

One pitch · 27 hours RFP-to-ship · 47 min of senior time

Pitches that used to need a week. Done before lunch tomorrow.

Manual cycle: 5 days, 8h of senior assembly, deadline always tight. With maxodev: 27 hours start-to-ship, 47 minutes of senior time, buffer to spare. Click through one pitch as it unfolds.

Draft v0 · not startednorthwind-capital · RFP-2026-114

Pitch to Northwind Capital — RFP-2026-114

RFP received · intake engine triggered · senior unblocked
Senior strategist · in meeting · 0 min spent

Example workflows

Four we've shipped. Your real workflow could be the fifth.

These aren't products on a price list. They're examples of what an audit-first rebuild looks like — built end-to-end on the client's real data, with metrics agreed before we start. We've also shipped automations for client onboarding, scope creep tracking, creative testing, competitor research, AM assistance and meeting-to-task routing. If your bench is leaking somewhere else, that's what the audit is for.

01
Pitch engine
4-week pilot

First-draft pitch decks from your own case studies.

  • Tone-aware draft from your last 100+ decks
  • Sector-tagged case retrieval, no generic boilerplate
  • Slides API export, ready for senior review
Before
8 h
After
45 min
02
QBR & reporting engine
6-week pilot

Client narratives drafted from the warehouse, not from memory.

  • Looker / BigQuery → 12-slide draft with citations
  • Tone profile per client account
  • Strategists edit, don't write
Before
6 h
After
28 min
03
Brief & scope engine
8-week pilot

Intake that turns any brief into a scoping doc in 90 minutes.

  • Requirements + risk extracted from Word / PDF / email
  • Effort estimates calibrated on your own timesheets
  • Routed to discipline leads with 2–3 confirm questions each
Before
3 d
After
90 min
04
Performance pulse engine
4-week pilot

Anomaly response that beats your Slack reply time.

  • 24/7 monitoring of CPA, ROAS, CTR and spend pace across Meta, Google, TikTok and DV360
  • Alerts arrive with a root-cause hypothesis (creative fatigue, audience overlap, bid issue) and a suggested next action
  • Catches the spike Monday at 7am — not Friday from a worried client email
Detect
2–3 d
After
<20 min

How a pilot looks

Working prototype in weeks. Pilot decides whether it pays.

Audit-first. We don't sell a retainer. We sell a 6–8-week experiment with metrics agreed on day one — and walk away if the numbers don't move.

Week 1

Audit

We sit with one workflow end-to-end. Real briefs, real decks, real reviewers. Output: a written audit and a go/no-go for a prototype.

Weeks 2–3

Prototype

Working prototype on your real data. No mockups. By the end of week 3 the senior reviewer is using it on a live piece of work.

Weeks 4–6

Pilot

Roll into one team, one client portfolio or one pitch cycle. We measure the metrics that matter — senior hours, throughput, win rate.

Week 7+

Decision

Pilot review. Numbers on the table. Roll out wider, kill it, or fold the lessons into a different workflow. No long-term retainer locks.

Who builds it

Two founders. Strategy and engineering, same room.

A senior pair, not an agency. We do the audit, the prototype and the pilot ourselves — no offshore handoff, no junior staff learning on your data. The cost of a pilot is roughly one month of a senior in-house AI hire, and you get a working system at the end of it.

Andrew Bulgakov

Andrew Bulgakov

Co-founder & CTO

10+ years building production AI systems — retrieval, agents, and the brittle glue between LLMs and real workflows. Has shipped automation pipelines used by enterprise procurement and operations teams. Turns “wouldn't it be cool if” into “shipped on Tuesday”.

Retrieval & RAGAgentsProduction MLData infra
Alex Hrechenko

Alex Hrechenko

Co-founder & Strategy

8 years across engineering, product and agency partnerships. Scoped 40+ AI workflows for B2B and agency teams — and walked away from twice as many that wouldn't pay back the budget. Believes senior afternoons are the most strategic thing AI can hand back to an agency.

Workflow auditsAgency operationsPilot scopingStakeholder mgmt

Field notes · agencies

Five workflows, five agencies, one pattern: senior margin back on the line.

Client names redacted under NDA. We share anonymized workflow screenshots and before/after artefacts on the audit call.

Who we work with

We're choosy on purpose.

Half the value of audit-first is saying no fast. Here's exactly who this is for — and who should pick a different shop.

Fit

  • 15–100 people. Independent or PE-backed agency.
  • EU, UK or US — we work in your business hours.
  • Doing real client work: performance, creative, content, integrated, production.
  • Margin pressure is real and you can name the two workflows that hurt most.
  • Leadership wants AI to free senior time, not to replace junior hires.
  • Willing to put real briefs, decks and reports into a 1-week audit.

Not a fit

  • Looking for generic AI advice or a chatbot demo.
  • Sub-15 people — not enough volume to justify a build yet.
  • Want a ChatGPT prompt training day or a workshop deck.
  • Looking for a slide presentation about AI, not a working prototype.
  • Not willing to share real workflow artefacts during the audit.

What it costs

Fixed scope. Fixed price. Audit fee credited to the pilot.

No retainers. No headcount-based billing. Every engagement is scoped to one workflow with a number we send before you commit.

20-min margin audit

Free

One workflow walked end-to-end on a call. Honest yes/no on whether an AI rebuild can pay back in six weeks.

1-week paid audit

Low five figures

Written audit, data assessment, scoped pilot proposal with a fixed price. Fully credited toward the pilot if you proceed.

4–8 week pilot

Fixed price

One workflow shipped on your real data, with the success metric agreed on day one. Walk away if the numbers don't move.

After the pilot

No retainer

Roll wider, kill it, or hand it to your team. No headcount minimums, no monthly lock, no ongoing fee.

Questions we get a lot

Real answers, not deflections.

We already tried ChatGPT and a few AI tools. How is this different?

ChatGPT saves a junior three minutes drafting an email. We rebuild the eight-hour task itself — the senior workflow underneath. A prompt template lives inside a person's head; a workflow we ship lives inside your pitch process, your warehouse, your approval chain.

If you're getting marginal time savings from off-the-shelf tools, the leak is upstream: in how briefs arrive, how decks get assembled, how reports get written. That's what we audit.

Will this replace our junior team?

No — and we won't take on engagements where that's the goal. The point is the opposite: senior time stops disappearing into assembly, juniors stop being review-bottlenecked, the agency does more pitches and better work, not the same amount with fewer people.

Every workflow we ship has explicit human-in-the-loop review steps. Senior judgement is the moat we're protecting, not replacing.

Our data is messy and lives across Drive, Notion, Frame.io, three CRMs. Will that work?

That's the normal case. Every agency we've worked with has the same shape: a Drive that's a graveyard, a Notion that's half-stale, a Frame.io with two years of feedback nobody can find. We don't make you clean it first.

Week 1 audit includes a data audit. We tell you what's usable, what's noise, and exactly how much cleanup the workflow actually needs (often less than you'd guess).

What about IP, client data, GDPR, US data residency?

EU-resident infra by default, US-resident on request. We use enterprise model endpoints (no training on your data), private vector stores, and SOC2-aligned vendors. Per-client data isolation is the default, not an upsell.

If your largest client has a specific GDPR or DPA requirement, send it over before the audit. We've answered the standard ones.

What does this cost?

Summary is in the pricing section above. Pilot pricing scales with workflow complexity, not headcount — we send a specific number after the 20-min call or after the 1-week audit, with a hard scope and a metric.

No retainers, no ongoing minimum.

Why not just hire an AI engineer in-house?

At 100+ people with internal AI engineering you probably should — and we'll tell you so during the audit. Below that, the math doesn't work: a senior AI engineer is a £180k+ salary and a 6-month hiring slog before they ship anything agency-shaped.

We're a two-person shipping team specialised in agency workflows. You get the equivalent of a senior pair for the cost of a pilot, and we hand you a working system at the end.

One audit · one workflow · no retainer

Book a 20-minute margin audit.

Bring one workflow that's eating senior time. We map it end-to-end and tell you honestly whether an AI rebuild pays back in six weeks — or whether the leak is somewhere else. No deck, no pitch, no follow-up unless you ask.

  1. You name one workflow eating senior time.
  2. We map the current path, volume and senior hours.
  3. You get an honest answer on whether it can pay back in six weeks.
  4. If not, we say no and tell you where the leak probably is.
Replies within one working day · EU, UK & US time zones