The Procurement Gap

Thirty per cent. That was the share of substantive Kenyan equity that any company providing ICT services in Kenya was required to maintain. This was the only regulatory instrument that structurally ensured Kenyan ownership in the technology sector. On 22 August 2023, the Cabinet Secretary for ICT signed Gazette Notice No. 11079, deleting the provision in its entirety effective immediately. No phase-out, no transition, no replacement mechanism.

In the same period, the same government was drafting the National AI Strategy 2025–2030, which calls for reviewing public procurement regulations to prioritise local AI solutions. Five months after the Strategy’s launch, the Judiciary’s AI Adoption Policy Framework also announced targets for AI deployment across case management and legal research.

One hand writes the strategy. The other signs the gazette. Neither talks to the other.

A new company now exists that these instruments describe. It is small. It is Kenyan. Its flagship product is a legal AI research assistant grounded in the full Kenyan legal corpus and designed to answer questions about Kenyan law. The product is live and gathering users. As I steer the strategy of this company, the hardest question I have been asked by friends and critics alike, and with increasing frequency, is whether the procurement system through which government institutions acquire technologies such as this one can deliver a contract to a small, nascent and unknown company such as ours.

In Kenya, public procurement accounts for sixty per cent of the Government’s annual budget. This typically amounts to between ten and thirteen per cent of GDP. Every directive in Kenya’s AI Strategy, every clause about local preference, and every provision about domestic capability, must eventually pass through this procurement system to mean anything at all. Policies that cannot survive contact with procurement are policies that exist only on paper.

A system of this scale also invites scrutiny. Kenya’s procurement system has been assessed, audited, reviewed, and benchmarked by its own regulatory authority, by the IMF, by the OECD, by the United States Trade Representative, and by many other institutions, all within the past four years. Their findings converge. And year after year, nothing changes.

The Silence

Suppose a small Kenyan tech company goes looking for a government contract. It searches the Public Procurement Information Portal. It monitors gazette notices, watches for announcements from the procuring entities whose needs it can serve.

What it finds, too often, is silence.

The 2024 MAPS assessment of Kenya’s public procurement system, published by the Public Procurement Regulatory Authority itself, surveyed sixty-three of the country’s major procuring entities. At least twenty-seven per cent did not publish their tender notices. One in five contracts was concluded without appearing in any procurement plan. These were, by design, transactions that were never intended to be open.

Even when tenders are visible, the architecture selects against the companies the national AI Strategy claims to favour. Persistent late payments from procuring entities threaten the survival of all small and medium enterprises. The MAPS assessment is precise: these delays also “create incentives for fraud and corruption in the form of facilitation payments for fast-tracking payment of invoices.” Thirty-five per cent of private sector respondents also stated that none of the enabling conditions for participation were met. Not some. None. The Kenyan procurement system therefore not only disadvantages small companies, but also metabolises their vulnerability, converting the distance between invoice and payment into a revenue stream for persons positioned to bridge it.

The Access to Government Procurement Opportunities (AGPO) programme was designed to reserve thirty per cent of government procurement value for women, youth, and persons with disabilities. In a decade of operation, utilisation has never exceeded seventeen per cent. This is barely half the thirty per cent target, and what it delivers, unfortunately, consists disproportionately of low-value contracts: the kind of work no “big” or connected firm would bother competing for. AGPO’s architects called it affirmative action. Its record is closer to affective appeasement.

The Absence of Consequence

Regardless, nothing really changes. The sharpest evidence of this is not in any report. It is in the hospital still processing claims by hand because the digital system it was promised never arrived. It is in a courtroom where a lawyer’s submission cited cases and statutes that do not exist. It is in a procurement office where a local company was identified, a solution proven, and yet the contract went to a firm whose principal qualification was knowing whom to call. It is in every quiet injustice, painfully felt but lying unseen, as the system moves on but the victims cannot.

The conventional reading of this by Kenyans is dysfunction. They say the system does not work because the system cannot work. The system is either too large and too fragmented, or too political and under-resourced to police itself. Procuring entities number in the thousands, and many of them are too small to employ a single qualified procurement professional.

However, the evidence suggests a reading far less comfortable. Any system that has been diagnosed by four independent institutions across four years and altered nothing in response is not a dysfunctional system. Dysfunctional systems cannot absorb diagnosis endlessly without submitting to a cure. This is a system that has been designed precisely that way, and it is functioning as its creators intended.

Consider the regulator. The OECD’s 2026 Peer Review of Competition Law and Policy in Kenya found that the Public Procurement Regulatory Authority has never imposed a single sanction for a breach of Kenyan procurement law. Not for bid rigging, not for non-compliance with transparency obligations, not for any violation of the rules governing how public money is spent. Is a regulator that has never sanctioned anyone a regulator?

Consider the transparency requirement. The International Monetary Fund made beneficial ownership disclosure a condition of Kenya’s loan programme in 2022. Procuring entities were required to disclose who owns the companies winning public contracts. This was a structural benchmark, meaning disbursements were contingent on compliance. By October 2024, barely half of companies on the register were compliant.

The IMF had set the condition. Kenya accepted the condition. The PPRA was tasked with enforcing the condition. And the system absorbed the condition without changing, perhaps because someone forgot to mention that the PPRA does not possess sanctioning powers for non-compliance, it issues letters. The benchmark was therefore met in form and defeated in substance: revised tender documents were adopted, a portal was created, letters were sent, and half the register still did not comply. This is not dysfunction. The system was designed to produce the documentation of accountability without following through with its practice.

The Verdict from Outside

In March 2026, the United States Trade Representative published its National Trade Estimate Report. The Kenya chapter documented what Kenyan institutions had been reporting for years, but in language that operated outside the closed loop of technical assessments and structural benchmarks. American firms reported “direct and indirect requests for bribes from multiple levels of the Kenyan Government.” Corruption was “widely reported to affect government procurements at the national and county levels.” Foreign firms, some without proven track records, had won government contracts “when partnered with well-connected Kenyan firms or individuals.”

None of this was new to Kenyans, or anyone who has worked with the Kenyan Government. The IMF had said it in 2022. The OECD confirmed it in 2026. Kenya’s own procurement regulator documented it in its own assessment. However, the USTR report landed where the others had not, because it was not written for the apathetic local system. It was written for an American audience, in the blunt diction of self-interest, and published where Kenya’s own accountability architecture could not contain it, and it landed accordingly.

Our small Kenyan tech company is yet to bid for a government contract. One day it will. When it does, it will enter a system whose own regulator has never sanctioned anyone, whose transparency obligations remain half unmet, and whose only structural protection for local technology ownership was deleted by gazette notice before it was founded, in the same season the AI Strategy was being written.

We know that the gap between Kenya’s AI ambitions and its procurement reality is not an oversight. It is maintained intentionally by opacity, by the absence of enforcement, by the quiet removal of protections, by a system that rewards proximity to power over proximity to competence. Most durably, by the production of reports and benchmarks and assessments that create the appearance of progress while the underlying architecture remains untouched.

We all know what such a system does to small companies, especially those with technical capability but no political connections. But we build anyway, because needs do not pause while institutions deliberate over whether to enforce their own rules. Citizens who cannot afford a lawyer will not wait for the procurement system to repair itself. When one hand builds and the other destroys, the ruins belong to both, and so the hand that builds must never stop building.