What will you do with infinite software?

PLUS: The hidden costs of agentic AI and how governance fixes them

From the aibl team

Over the break I had a chance for a virtual sit down with Stefan Tornquist to talk about trends for 2026. Stefan is an ex-Econsultancy colleague and our research partner at Executive Summary RSI.

The midmarket isn’t typically interested in trends that look out too far or too high to be practical in the near term. So, after some reining in, we focused on a macro trend that will increasingly affect every company in the midmarket; as AI improves in its ability to write code, software development shifts from scarcity to abundance.

As an experienced cynic about innovation and promised speed, you may contest the premise and that’s fair enough. But have a read of how Boris Cherny, who led the creation of Claude Code, sets up and uses the system, and you’ll get the picture. In essence, one senior dev can build a team of independent or parallel iterations of their top end model of choice, churning out, testing and debugging code.

Until now, our development needs have always far outpaced our capacity to build. That’s meant that an enormous amount of thinking, planning and architecting had to go into every step. We never wanted to waste a precious resource or make a misstep that would cost real time and money. What would it mean if a misstep meant a day of reworking or a week of completely replacing?

We’ve called data the ‘new oil’ for twenty years now, and like oil, data isn’t worth anything without extraction, refinement, distribution and application…that’s software in the analogy. The bottleneck to any company’s digital transformation, innovation and growth is less the lack of data than the software to use it.

When I’ve broached this topic with other business leaders, they immediately get a gleam in their eye about the prospect of ending the forever ‘buy vs. build’ debate. In the medium to long term that’s probably true, but I doubt that the first order of business will be to build a replacement CRM. It will take time for contracts, stacks and familiar processes to change.

I expect (and already see) the early shift to be in building lightweight internal tools for specific departments, building a focused application instead of purchasing a bolt-on or point solution.

In the last two weeks I’ve seen two examples that bring this to life.

The first is a dynamic commission engine that ingests data from the CRM and accounting systems, and uses natural language to compare against the comp plan, giving every commercial person a real-time dashboard showing the what and why of each payment, with a dispute button to flag the item for an agent (then a human) to investigate.

The second is an invoice checker that examines every inbound PDF invoice, reads the unstructured data and compares it with the original contract or rate card. Not only does it instantly approve 95% of invoices and ship them along for processing, it flags the 5% where something doesn’t line up.

These are ‘boring’ examples of powerful improvements that save time and money. But, they’re something that would have teetered on the line of ‘nice to have’ not long ago. Today, instead of a months-long dev cycle and a price tag with a comma or two, applications like this can be developed by one dev and AI over the course of days or weeks.

We’ll return to this trend, but for now, let me know what your organisation is doing in this vein. richard@aiblmedia.com

---

The Brand Constitution: How to Automate Brand Governance

The Hidden Editing Tax

A mid-market brand used an agent to generate 500 product descriptions, finishing the first draft in under an hour.

A junior marketing manager was asked to review the copy. By the third description, problems were obvious.

‘Delve into a world of comfort.’

‘Unleash the power of sustainable living.’

What should have been a quick check became a week of manual editing. The manager fixed the same errors on a loop: stripping clichés, tightening tone, and adding back mandatory details.

The agent hadn’t saved work; it had just shifted it to expensive human oversight.

The Prompt Mirage

The team believed a prompt could guarantee on-brand content. Managers saw coherent text and assumed the job was done. It failed for two reasons.

• Brand knowledge is implicit. The real voice of a brand lives in people’s heads. It is shaped by unwritten rules, past decisions, and judgement built over time. None of this exists in a prompt unless made explicit.
• Scale makes this worse. The same mistake across 500 descriptions becomes an editing tax. Humans stop editing and start policing output.

Read more.

---

NEWS

We’re still collecting stories from the field, what worked, what didn’t, and what surprised you. If you’ve got a case study that belongs on the aiblLIVE stage, get in touch with via John@aiblmedia.com.

1. Confidence Is Back But Capacity Isn’t, Dec 31 – Lloyds’ December Business Barometer puts UK business confidence at 47%, with economic optimism at a four-month high as post-Budget uncertainty eases. But staffing plans have softened. Hiring intentions slipped to a three-month low even as 62% of firms expect stronger output next year. Wage pressure is still there too, with 18% anticipating pay growth of 4% or more, while own-price expectations edged lower again. This is when agents start to look sensible. Not as a grand strategy, just a way to take repeatable work off teams so growth doesn’t just create bigger backlogs.
2. Agents Expose the Mess in 2026, Jan 2 – A piece quoting senior execs from Avanade, Tech Mahindra and F5 says 2026 is when AI gets judged on outcomes, not demos. Many firms stalled in 2025 because they couldn’t prioritise which processes to tackle, data was scattered, and teams couldn’t describe workflows in a way AI could use. Agents need the real steps, including judgement calls and exceptions, so they force that clarity. The value isn’t headcount reduction. It’s helping teams solve problems, make decisions, and flag issues earlier. AI becomes real through APIs, so the plumbing matters. If you don’t know what’s connected, you don’t know where the weak points are. Agents won’t fix chaos, they’ll just run it faster, so map the work and define the task before you delegate.
3. HCM starts tracking agents too, Forrester (Predictions 2026) – The next question is governance. How do you track what work is done by people, and what work is done by agents? Forrester predicts the top HCM platforms (human capital management, essentially HR systems) will add ‘digital employee management’. The idea is that HR tech starts tracking AI agents alongside people, as more work gets done by role-based agents running across multiple systems. They also flag mid-market firms as early beneficiaries, because they’re feeling the capacity squeeze first. This shifts agentic AI from a tech purchase to a visible workforce strategy. If agents are tracked in HR systems, you can view human and digital capacity in one place. It gives leaders a clear, tradeable choice: do we hire to close this gap, or do we formally roster an agent?
4. 2026 AI Trends in Mid-Market Financial Management, Dec 11 – And that control shift is already happening. A Citizens Bank survey of 134 US mid-market firms shows 82% plan to increase AI investment over the next five years. Realised ROI (35%) is closing in on the ‘success’ target of 41%. Use of external AI partners fell from 64% to 58% year on year, with teams moving from outsourcing to building capacity in-house. For agents, 82% plan to implement in 2026, focusing on fraud and security. Don’t over-read the adoption rates since the sample excluded non-AI users. The clearer signal is the drop in outsourcing as AI moves closer to core operations.

---

PRODUCT SPOTLIGHT OF THE WEEK

This week we’ve been looking at Search Atlas, an SEO platform for mid-market teams fed up with audit tools that only generate a longer to-do list. Its AI agent, OTTO, is built to close the gap between spotting issues and getting fixes live.

Add a pixel to your site and it pushes proposed on-page changes to your dashboard for approval, covering meta tags, schema, canonicals, and internal linking. It’s a practical answer to the SEO backlog that sits untouched because marketing has no spare headcount and dev time is rationed.

Pixel-driven fixes won’t replace proper server-side work, and changes only persist while the pixel stays in place. Keep strategy in-house, and use OTTO to speed up the repetitive on-page changes once you’ve reviewed them. Start with one property, then expand if the impact holds and nothing breaks.