On February 25, 2026, FBI agents served search warrants at two locations: the San Pedro home of LAUSD Superintendent Alberto Carvalho, and LAUSD headquarters. Within 48 hours, the board voted to place Carvalho on paid leave. By May, the district’s top academic chiefs had departed.
This wasn’t about a stolen laptop or a vendor overcharging for textbooks. This was about a $6.2M AI chatbot contract that went wrong — and a federal probe that’s still active as of this writing.
The timeline should terrify every school board.
February 2024: LAUSD awards AllHere Education a $6.2M sole-source contract for a district-wide AI attendance chatbot called “Ed.” No competitive bidding. No RFP. No price discovery.
Mid-2024: AllHere furloughs most employees. LAUSD shelves the chatbot. The district had paid roughly $3M of the $6.2M before the company imploded.
September 2024: AllHere files for Chapter 7 bankruptcy in Delaware.
November 2024: Founder and CEO Joanna Smith-Griffin — a Forbes 30 Under 30 alum — is indicted on charges of defrauding investors of nearly $10M. Securities fraud. Wire fraud. Aggravated identity theft.
March 2026: Investigative reporting reveals AllHere’s CEO met privately with LAUSD leaders months before the board approved the deal. Miami-Dade, it turns out, had awarded a separate $1.89M contract for similar services.
The LAUSD board voted 5-2 to approve the AllHere contract. The two dissenting members cited lack of vendor due diligence and single-source concerns. They were right.
Here’s the question every board member should be asking themselves right now: Could this happen on my board?
If your board doesn’t have a framework for evaluating AI vendors — if you’re approving technology contracts the same way you’ve always done it — the answer is yes.
Here are the five questions you need to be asking before the next AI contract lands on your agenda.
Question 1: Why is this a sole-source deal?
AI vendors love to pitch their products as unique. “There’s nothing else like it.” “We’re the only ones doing this.”
That’s not a feature description. That’s a sales script.
Sole-source contracts eliminate price discovery, feature comparison, and the financial vetting that competitive processes provide by design. LAUSD’s AllHere contract was sole-source. The board never saw alternative bids, never evaluated competing products, never had a basis to know whether $6.2M was reasonable or ridiculous.
The board action: Require competitive procurement for all technology contracts exceeding $100K. If the administration insists on a sole-source exception, require a documented emergency justification with a timeline for competitive re-bid built into the contract terms.
The ESB framework: This is Document 9 — Effective Professional Services Management — in action. The Purpose phase asks: Why are we buying this, and what student outcome does it serve? The Selection phase asks: What criteria are we using, and do they include competitive comparison? If you can’t answer those two questions, you’re not ready to vote.
Question 2: Can this vendor survive the school year?
AllHere was a startup with a single major client — LAUSD. When the contract ran into trouble, there was no revenue cushion. No diversified client base. No runway.
The company’s financial instability wasn’t public at the time LAUSD signed the contract. But standard due diligence — audited financials, client concentration analysis, runway assessment — would have flagged the risk.
Startups fail. It’s not a moral failing. It’s a statistical reality. When a school district pins a multi-million dollar initiative on a vendor that might not exist in 18 months, that’s not innovation — it’s negligence.
The board action: Require audited financial statements, client concentration disclosure, and evidence of at least 12 months of operational runway as part of every technology vendor RFP response. If the vendor can’t or won’t provide it, that’s your answer.
The ESB framework: Document 11 — Effective Risk Management — directly addresses this. The Goals/Guardrails framework asks: What happens if the vendor fails? If your answer is “we’ll figure it out,” you haven’t done risk management.
Question 3: Did the board actually oversee this, or was it a staff decision with a board stamp?
In the AllHere case, the superintendent and district administration negotiated and awarded the contract. The board voted. But the substantive discussion — the technology’s rationale, the evaluation criteria, the risk profile — never happened at the board level.
That’s not board oversight. That’s a rubber stamp.
Your board’s governance role is setting guardrails and monitoring execution. When a technology contract bypasses board-level scrutiny — when the conversation happens in closed-door meetings between the superintendent’s office and the vendor — the guardrails disappear.
The board action: Adopt a guardrail requiring board review and approval of all technology contracts over $250K. The administration should present vendor selection criteria, competitive alternatives considered, and a written risk assessment. No exceptions.
The ESB framework: Document 8 — Effective Policy Leadership — gives you the architecture. The Delegation domain defines what the superintendent can do without board approval and what requires a board vote. If your policy doesn’t draw that line for technology contracts, it needs updating.
Question 4: Did we pilot this before scaling it?
LAUSD committed district-wide before validating whether the chatbot worked. No pilot. No success metrics. No go/no-go gate.
The AI startup pitch is always about potential. Potential is not proof. A phased deployment — pilot in 3–5 schools for one semester with measurable outcome criteria — would have caught technical, adoption, and vendor viability issues before the full $6.2M commitment.
Your board has the authority to demand proof before scale. Use it.
The board action: Require a pilot phase with clearly defined success metrics for any novel technology deployment. Full rollout must be contingent on pilot results, with an explicit go/no-go decision by the board.
The ESB framework: Document 10 — Effective Research Analysis — gives you the PEARLS framework for evaluating vendor claims. Is the evidence from a peer-reviewed study or a vendor white paper? Does it show causation or just correlation? PEARLS is how you tell the difference between a real solution and a well-funded sales pitch.
Question 5: Did we define the problem before we picked the solution?
Here’s the question that catches most boards — because by the time it lands on your agenda, the solution has already been chosen.
The AllHere chatbot was marketed as a transformative attendance tool. Maybe it was. Maybe it wasn’t. The board never had a framework to evaluate whether the claimed benefits were realistic, whether the technology was appropriate for LAUSD’s context, or whether a less expensive — or non-AI — alternative would have achieved the same outcome.
You might be surprised how often the answer to “what problem are we solving?” turns out to be: “we don’t actually know.”
The board action: Institute a requirements-first approach. Define the problem and the desired student outcome before any technology solution is presented. Require the administration to present at least one non-technology alternative alongside every AI vendor proposal.
The ESB framework: Document 3 — Effective Budget Alignment — is your anchor here. Before approving AI spending, verify that the line item ties to a specific Goal. And make sure the budget reflects the total cost of ownership — licensing, training, infrastructure, support, and eventual decommissioning — not just the first-year subscription fee. A $6.2M contract starts to look different when you ask which Goal does this serve?
Now What? A Three-Step Meeting Sequence
You’ve read the questions. Now here’s the work.
This month. Add AI procurement guardrails to your next board agenda. Adopt a policy requiring competitive bidding for technology contracts over $100K and board review for contracts over $250K. Your administration needs to know the new rules before the next vendor walks through the door.
Next month. Audit your current technology contracts. Which ones are sole-source? Which vendors have shaky financials? Which contracts lack a student data privacy clause? You can’t fix what you haven’t inventoried.
This quarter. Schedule a board workshop on AI governance literacy. Bring in an independent expert. Walk through the vendor evaluation rubric. Role-play an AI contract approval with the PEARLS framework. The goal is not to make every board member a technologist. The goal is to make sure no board member votes on something they don’t understand.
Don’t wait for the FBI to find the gap in your governance.