You need the thing built.

Then borrow a marketing lead part-time. A fractional CMO who sits inside your team: your positioning, your plan, your budget, your hiring brief — and someone in your Monday meeting who owns the number. Two days a month, or two days a week.

Reaching customers and growing sales is the single most-reported operational challenge (US Federal Reserve, 2026 Report on Employer Firms, n=6,525). 70% of owners spend under five hours a week on marketing (Fiverr, n≈6,000); confidence fell from 27% to 18% in a year (Constant Contact, n=2,500).

We automate the chase. Invoices out on time, reminders that go without you writing them, reconciliation that stops being a Sunday job, and a dashboard that tells you who is late before payroll does.

59% of US small businesses have invoices 30+ days overdue, up from 47% a year earlier; average unpaid balance $17.7K; 39% say a single late payment made payroll or bills hard to cover (Intuit QuickBooks Late Payments Report, 2026). Invoices are now paid 9.0 days late on average and worsening (Xero, observed data, 32,000+ firms).

That is the normal outcome, and the reason is documented: the tools do not remember your preferences, do not learn from corrections, and do not know how your approvals or your data actually flow. We start from your workflow rather than from a model, and we only keep what survives a month of real use.

53% of small firms say their operations would be unaffected if AI tools disappeared tomorrow (Xero, n≈1,100). Buyers in MIT's 2025 study named the cause directly: “Most vendors don't get how our approvals or data flows work.”

We train your team on your own work rather than on toy examples: a half-day for the whole office, then role-by-role sessions for the people who will use it daily, and a written playbook of the prompts and workflows that actually stuck.

73% of small businesses say training and implementation support is what they need most; only 14% have fully integrated AI into core operations (Goldman Sachs 10,000 Small Businesses, n=1,256, 2026).

Fair question, and usually the honest answer is yes. We build assistants that run against your documents, your policies and your products, inside a boundary you control, with a written statement of what leaves your systems and what never does.

Data privacy and security is the single most-cited barrier to AI adoption, at 36%, ahead of accuracy at 29% (Xero, 2025).

That spreadsheet is a specification. It is the part of your business no product on the market models properly — so we build that part, and leave the rest to the tools you already own. Usually smaller than you fear, and it starts as a v1 you can test in weeks.

71% of firms still run spreadsheets alongside or instead of a CRM (Zoho, n=5,149); 58% of small businesses now use four or more separate tools and a third use six or more (US Chamber C_TEC, 2025).

So we fix what it reads. Positioning, the words, the structure and the site itself: not a moodboard, but a working brand system with tokens, type and components, built so both a person and a model can tell what you do in one pass.

90% of small businesses plan to invest in their website in the next 12 months and 58% want a redesign (Clutch, n=406). Worth knowing: owners spend on this without calling it a problem — NFIB's forced ranking puts advertising cost-effectiveness at #46 of 75.

The gap is almost never the model. It is ownership, data readiness and the route to a permanent budget line. We take one pilot, put the plumbing and the governance under it, and get it into production — then use it as the template for the next four.

The average organisation scrapped 46% of AI proofs-of-concept before production, and 42% of businesses scrapped most of their AI initiatives in 2025, up from 17% in 2024 (S&P Global Market Intelligence, n=1,000+). Only 16% of AI initiatives have scaled enterprise-wide (IBM IBV CEO Study, n=2,000).

Because it is. We treat integration as the product rather than the afterthought: the connections, the contracts between systems, the identity and the audit trail first — then the agent, which is now the easy part.

The average enterprise runs 957 applications and only 27% are connected; 86% of IT leaders agree that without proper integration, AI agents add more complexity than value (MuleSoft Connectivity Benchmark, n=1,050 IT leaders, 2026). Salesforce owns MuleSoft, so read the framing accordingly — the ratio has held for years.

Then the security review is the project. We do that half first — data boundaries, model inventory, access, logging, a written risk frame — and hand your reviewers something they can approve rather than something they have to interrogate.

Regulatory compliance became the #1 barrier to AI adoption, rising from 28% to 38% in a single year (Deloitte, n=2,773). Roughly two-thirds of organisations now name security and risk as the top barrier to scaling agentic AI, ahead of both regulation and technical limits (McKinsey State of AI Trust, 2026).

So we build the evidence in rather than bolt it on: model inventory, decision logs with retention, documented data lineage, human-review points, and the paperwork that maps to the regime you actually fall under.

The EU AI Act's high-risk obligations — automated logging, six-month retention, technical documentation, conformity assessment, human oversight — bind from 2 August 2026, and more than half of organisations do not yet hold a systematic inventory of their AI systems (Cloud Security Alliance research note).

We have rebuilt programmes around smaller, later and less certain money — restructuring what gets delivered, finding the co-funder, and writing the case that keeps the rest of it. It is unglamorous and it is most of what we do for institutions right now.

US NIH awards fell 29% and NSF awards 50% in-year, with roughly $17bn of NIH grants disrupted. Nearly 70% of nonprofits report cuts from at least one funding source (Center for Effective Philanthropy, n=408); 45% of English higher-education providers are projected to run a deficit in 2025-26 (Office for Students).

Not quite half — 44.3%, and it has not improved in twenty years. We automate the recurring parts: proposal assembly, post-award reporting, compliance evidence and the data collection your funders keep asking for in a slightly different shape each time.

Faculty spend 44.3% of federally funded research time on administration; 62% say the burden seriously compromises research productivity and 31% say it discourages them from submitting federal proposals at all (Federal Demonstration Partnership Faculty Workload Survey, n=11,167 principal investigators).

Everyone's are. We write the policy with the faculty rather than at them, build the assessment guidance students actually asked for, and put the practical training underneath so the policy is enforceable.

88% of students and 77% of faculty now use AI, but only 31% of faculty were meaningfully involved in setting their institution's AI policy and 57% of students say they lack adequate guidance on AI in assessment (Digital Education Council Global Survey 2026, n=45,398 across 35 countries). Only 14% of provosts report comprehensive AI governance (Inside Higher Ed / Hanover, n=478).

Usually because the programme was an event rather than a curriculum. The evidence is blunt about this: money and desk space change nothing on their own — the structured teaching is what moves the outcome. So we build the teaching, the mentor pool, the documentation and the contributor path, and we run them long enough to matter.

In the cleanest causal study available (Gonzalez-Uribe & Leatherbee, Review of Financial Studies, on Start-Up Chile), cash plus co-working produced no measurable improvement — the entrepreneurship schooling component drove the gains. Matched-sample studies of incubators find null or negative effects on survival (Schwartz, J. Technology Transfer, n=371 pairs).

We turn your artworks, archives, catalogues, curricula and research into assets that keep earning — licensing, digital editions, immersive versions, documentary. Consent, ownership and royalty splits are written into the contract before anything is built.

Our own reference case, Design Baltimore: Gateways with HarlemCLX, is a $450,000 cultural infrastructure engagement modelled over fifteen years — see the scenario range in section 06.

Documentary, not advertising. Long-form series, live capture, editorial strategy and the distribution behind it. Beyond the Code has run since 2022 across four continents, narrated by Macy Gray.

A documentary's useful life is measured in years; a press release's in hours. This is the one line on the page we assert from practice rather than from a survey.

We are already there, and the introductions are the product. Embedded practitioners in Nairobi, Addis Ababa and Chennai, with hubs in Singapore, New York, Los Angeles and Munich.

Kenya took $984m of Africa's ~$3.2bn startup funding in 2025 — 31% of the continent and the top market (Africa: The Big Deal). India recognised 55,200+ new startups in FY2025-26 alone, and Chennai has the highest concentration of academic incubators of any Indian city at 82% (DPIIT; India Incubator Kaleidoscope, IIT Madras CREST / NSRCEL).