Khalil Dimachkie

AI agentic startup Gradient Labs has raised $13 million in a Series A funding round. Gradient Labs is reinventing customer support for regulated industries such as financial services. Its AI agent is purpose-built to handle complex, compliance-heavy customer operations, going beyond simple frontline chatbots. The platform can resolve up to 90% of queries with a 98% quality assurance pass rate by understanding company-specific processes, navigating ambiguity, and reasoning before answering, all while reducing costs. The round was led by Redpoint Ventures, with participation from LocalGlobe, Puzzle Ventures, Liquid 2 Ventures, and Exceptional Capital. The new capital will enable Gradient Labs to expand its technology, marketing, sales, and customer success teams, and increase its investment in R&D, with a focus on geographic market expansion and adding voice capabilities. Congratulations to co-founders Dimitri Masin, 🚀 Neal Lathia, Danai Antoniou, and the Gradient Labs team. Tech.eu: https://lnkd.in/dCHxDkZ2 #FinTech #AI #AgenticAI #CustomerSupport #RegTech #SeriesA #Funding #UKTech

5

𝐃𝐞𝐥𝐚𝐲𝐞𝐝 𝐃𝐮𝐞 𝐭𝐨 𝐄𝐚𝐬𝐭𝐞𝐫 When I worked in the computer industry, one of the highlights of the week was the Software Development Review. People came in from all over - Scotland, the Midlands, London… and rural Berkshire - and each section head had to account for progress, or the lack of it, walking through milestones, exceptions, what had gone wrong, and what was being done about it. It was disciplined, tense, and high-pressure. No one wanted to be humiliated. It was chaired by a Director who, had he not gone into systems, might well have found a successful career in organised crime. He even had a Sicilian name that sounded like it belonged in organised crime. On one memorable occasion, a Development Manager was asked why his program had slipped. His answer was simple: “𝘋𝘦𝘭𝘢𝘺𝘦𝘥 𝘥𝘶𝘦 𝘵𝘰 𝘌𝘢𝘴𝘵𝘦𝘳.” There was a pause. The Director turned to me and asked, “𝘞𝘩𝘢𝘵 𝘺𝘦𝘢𝘳 𝘪𝘴 𝘵𝘩𝘪𝘴?” I told him. He nodded. “𝘛𝘩𝘢𝘵’𝘴 𝘸𝘩𝘢𝘵 𝘐 𝘵𝘩𝘰𝘶𝘨𝘩𝘵. 𝘞𝘦’𝘷𝘦 𝘩𝘢𝘥 𝘵𝘸𝘰 𝘵𝘩𝘰𝘶𝘴𝘢𝘯𝘥 𝘺𝘦𝘢𝘳𝘴 𝘵𝘰 𝘨𝘦𝘵 𝘶𝘴𝘦𝘥 𝘵𝘰 𝘌𝘢𝘴𝘵𝘦𝘳… 𝘢𝘯𝘥 𝘸𝘦 𝘴𝘵𝘪𝘭𝘭 𝘤𝘢𝘯’𝘵 𝘱𝘭𝘢𝘯 𝘢𝘳𝘰𝘶𝘯𝘥 𝘪𝘵. 𝘈𝘳𝘦 𝘺𝘰𝘶 𝘴𝘦𝘳𝘪𝘰𝘶𝘴?” It was funny - but it has stayed with me ever since, and I still wheel it out around Easter when I sense people are tempted to soften or hide bad news, because the underlying point is not funny at all. How often do we accept explanations like this in program reviews - things that sound plausible until you look at them properly, and then quietly disintegrate? Listen carefully to the language: qualifiers and modifiers usually reveal what hasn’t happened, and if the explanation is something the organisation knew about in advance, it isn’t an explanation - it’s a planning failure. Happy Easter! 🐰

11

1 Comment

Somerset NHS Foundation Trust to move the Somerset Cancer Register (SCR) to a cloud-based platform. There are 100 trusts which rely on the SCR to be able to track, monitor and manage patient cancer pathways, making it the backbone for cancer teams across England. “This launch represents a significant step forward in our commitment to modernizing cancer data workflows. By leveraging cloud technology, we’re ensuring that SCR remains a reliable, future-proof platform that meets the evolving needs of cancer services.Ruth McCarthyy, Operations Manager of the SCR,Somerset NHS Foundation Trustt. The trust has been working withMicrosoftt and specialist digital services consultancyAscentt to design the infrastructure and optimize the SCR application in the cloud, with go live planned for 6 October 2025. To learn more, read the full article: https://lnkd.in/gjzpUccM #Microsoft #Healthcare #AI

14

1 Comment

How to sell to the NHS? 🏥 Spoiler: it’s not about disruption, it’s about alignment. Last week I joined the National HealthTech Series in London, an insightful event bringing together healthtech founders, investors, and NHS leaders. 🚀 Here’s the main takeaway for anyone building in UK healthcare 👇 ✅ Understand exactly how the NHS works in your space. ✅ Build around their workflows, don’t try to replace them. ✅ Your product should fit seamlessly into existing processes. ✅ Deliver a step-change in outcomes. In healthcare (as in any business), success starts with understanding your customers’ pain points and nailing your value proposition. Thanks to Kosta Mavroulakis, FRSA and Jemma Cooper for organising and hosting such an inspiring event 🙌

9

The NHS Federated Data Platform (FDP) is rapidly emerging as a leading global example of effective digital health transformation, showing how data, used safely and responsibly at scale, can deliver faster, measurable improvements in clinical and operational outcomes.

10

Reposting a short post from James Woodward at the BCS drawing attention to the new 10 year plan for the NHS and specifically the role AI will play and the intention to train the workforce (no mean task given the size of the NHS). Would endorse and emphasise James' attention to this. My own doctoral research (built on with post doctoral research and consulting) looked at 150 UK organisations and what leads them to be more (or less) successful at the strategic management of information technologies (including AI). Not unexpectedly the UK's healthcare organisations in the study lagged most other sectors. The key to organisational success at technology is complex. But the study found that the skills and capabilities of staff with technology is one of the most significant enablers. As a consultant I have seen too many plans that mention skills but don't follow through well enough. My advice from my research and consultancy is that efforts on skills for AI cannot be over-emphasised. IMHO it is unlikely that efforts can ever be too high (there is a natural brake anyway from time available that can be freed up). This is not just applicable to the NHS plan - but to all sectors and globally. Hopefully, this part of the NHS plan is not token, but will get driven from the top teams across the NHS (and not just the training teams in the HR departments). Importantly, it will need funding, and it is worth remembering that investment in IT in the NHS has been incredibly low compared to other sectors for decades. This plan is a great first step. The direction of travel incontestable. Oh, and partnering with the BCS would be worthwhile too :-)

5

Just released some new heretic abliterated models for ComfyUI. I was really surprised with my first release of Gemma 3 12b v1 in just over four weeks it already had over 14,000 downloads. With the new release of LTX2.3 some people had requested the vision capabilities in Gemma 12b as newer ComfyUI nodes use them. We also have provided an NVFP4 version. > Blog update about the docker and how they were made - https://lnkd.in/gZ5uMA64 > Gemma 3 12b Heretic v2 for LTX2 and LTX2.3 - https://lnkd.in/gmn4q4iK > Qwen3 4b Heretic for ZImage and Klein 4b - https://lnkd.in/gZCiN7yK And check out the docker configuration used to make all of these here https://lnkd.in/gJN8F3_q

3

When we talk about interoperability, too often we frame it as a technical integration problem. But as I explored in my session at the British Computer Society earlier today, the reality is deeper and more strategic. 🔍 Application Design Debt is the hidden cost we pay when systems are built to trap data, treating integration as an afterthought. This fuels the Design Debt Cascade, UI-driven logic, data fragmentation, integration decay, and ultimately unusable data. The result? Interoperability efforts turn into archaeological salvage work — painstaking salvage operations to retrofit data from systems never designed with data integrity or exchange in mind. It’s not only costly, it slows down transformation and undermines trust in data. To break this cycle, we need Data-by-Design principles as the foundation for our systems of record: - Data-first architecture - Integration-native design - Metadata as a first-class citizen - Audit and lineage by default But technology alone is not enough. True interoperability and the vision of a trusted single patient record, demands equal investment in enterprise data management and business change, alongside people, processes and technology. This isn’t just about the next national strategy or programme, it’s about building the data foundations for our collective future: reducing clinician burden, enabling real-time care coordination, and unlocking trustworthy AI and population insights. 👉 My call to action: let’s stop patching around the edges and start designing/procuring systems built with data in mind from the ground up. Interoperability is not a bolt-on. It’s a choice we make at the architecture and procurement stage. Carol Munt Lee Rickles Ming Tang Ayub Bhayat Tito Castillo (FBCS CITP CHCIO) David Hancock Wendy Dearing Andrew Griffiths BCS Faculty of Health and Care 📸 From today’s BCS Faculty of Health & Care conference #HealthInformatics #Interoperability #DataByDesign #DesignDebt #SinglePatientRecord #DigitalTransformation American College of Healthcare Executives American College of Healthcare Architects American College of Health Data Management

38

9 Comments

In December 2025, NHS England quietly removed all open source policy pages from its website. No announcement.  Just 404 errors where commitments used to be.  In the same period, the same NHS solicited proposals for a Single Patient Record that requires vendor-neutral, open-standards architecture. The policy vanished. The need did not.  I have watched this pattern repeat for twenty-five years. Open standards mandates are politically convenient to announce and administratively convenient to abandon. The HHS AI Strategy requires FAIR data and open standards, with a compliance deadline of April 2026. What about 2027?  The question: if your interoperability depends on a policy page, what happens when someone deletes it?  In this piece, I argue that there are two kinds of openness. Mandated openness lives on a webpage and can be deleted.  Engineered openness lives in the data itself and cannot. When every datum carries its own constraints, its own semantic identity, its own provenance in W3C-standard formats, no policy rollback can strip that away.  The data IS the contract.  Full post: https://lnkd.in/erJfGD5h

40

9 Comments

Where software once ate the world, it is now starting to move it. AI is revolutionising how we discover drugs, yet the movement of the physical atoms of care has barely changed since the founding of the NHS in 1948. Highly trained clinicians are still being used as human middleware, chasing deliveries, moving samples and bridging disconnected systems. The UK is the best place in the world to build autonomous healthcare infrastructure, and the timing is now. Read the blog below: "Autonomously United Kingdom: Why the future of healthcare is looking up." https://lnkd.in/eJFutF9z

167

10 Comments

Earlier this year we entered the Innovate UK Agentic AI Pioneers Prize - a national competition for UK businesses building AI systems that take autonomous initiative across complex workflows. There was one winner per sector. We weren't selected. Our technical approach scored 8.7 out of 10. One assessor described the core thesis as compelling - that trade compliance should be "front and centre rather than a problem for later." What we're taking forward: → Clearer evidence standards - less describing, more demonstrating on real trade data → A sharper framing of where agentic AI surfaces recommendations and where professional oversight must sit → A more defined path from beta to production-grade deployment We're now in the Sie Ventures Foundations Programme, running our beta cohort, and moving toward our pre-seed raise. The competition didn't select us. The feedback made us better.

17

3 Comments

Good news is that our medical test results are accessible online and can be downloaded. Okay, so my HGB and RPC results show yellow, meaing that they're outside the normal range. Do I need to follow up on that? Do I care? You can upload your tests into a local DB that you control and only you have access to. Then you can use standard tools or use a tool like Codex or Claude Code to build your own medical portal, which is what I did. So I have long term trends and I also can ask questions about specific medical conditions I have, where I bring AI into the picture. I have a "roadmap" of new information sources I want to add and functionality and I work with AI daily to work through the backlog. I'm finding this a great way to increase my AI knowledge as well as hands-on skills.

2

AI in the NHS: Why Planning, People and Projects Matter More Than Hype Sorry to burst the bubble but leveraging AI in healthcare is not going to happen by magic or luck. For many who are aware, there has been a lot of talk about AI’s potential to transform diagnostics and reduce NHS backlogs. Though AI has its merits and potential, as I have always said, AI is simply another type of technology and like any digital tool, its success depends on strategy, planning, stakeholder engagement, and rigorous project management. A new study by UCL, the Nuffield Trust and the University of Cambridge confirms this reality. Despite a £21m investment to roll out AI across 66 NHS hospital trusts, 18 months later, a third of trusts had still not embedded the tools in clinical practice. The barriers were governance, procurement delays, ageing IT systems, workforce scepticism, and a lack of time for clinicians already stretched to breaking point. The findings reinforce five points I have consistently emphasised: 1. Throwing money and technology, in this case, at NHS backlogs, will not fix the problem. 2. AI adoption requires strategic thinking, structured planning and time to realise measurable outcomes. 3. Human potential must grow alongside technological potential. Staff must be trained, engaged early and carried along. 4. The injection of each new technology must fit into the existing enterprise architecture and align with or trigger an improvement of existing processes. 5. Governance, procurement strategy, outcome-driven business cases and dedicated project delivery teams are non-negotiable. This echoes a truth many overlook. Stakeholder engagement is not a tick-box task. It requires a real investment of time. Without it, we risk optimism bias, assuming AI will succeed just because it is new and exciting. The “as is” environment must be fully understood before change is introduced. The lesson for leaders is simple. AI is not a silver bullet. It will only deliver value when we treat it as part of a carefully managed transformation where people, process, strategy and governance are given as much weight as the technology itself. Knowledge Incorporated #BusinessOfProjectManagement #BusinessCase #TransformationLeadership #PortfolioManagement #DigitalTransformation #P3M #PortfolioDirector #ProjectManagement, #Resilience #CyberSecurity #Leadership #ai #nhs

15

4 Comments

This was mentioned by Lord Vallance at the WIRED Health Health conference this am … ➡️ 📈 💷 the UK government is launching a £500 million Sovereign AI fund aimed at boosting support for domestic AI companies. BTW - ScaleUp Group are successful founders (over £4bn in EV delivered), who have already helped 19 UK tech startups scale - with priceless scaleup leadership insights/experience plus personal funding to demonstrate “skin in the game” to founders and fellow investors. Story via ITPro … “While UK AI startups raised £6 billion in venture capital last year, the government said too many firms are failing to make the leap from breakthrough research to large‑scale commercial success. To drive industry growth, a new Sovereign AI Unit will act as a venture capital fund within the government, making targeted investments in UK AI firms. In addition to funding support, the unit will provide direct access to the UK’s fastest AI supercomputers, such as the Isambard-AI supercomputer in Bristol and Dawn in Cambridge, along with specialist R&D support and grants. The unit will also offer opportunities to engage directly with the government on procurement opportunities, along with help to navigate and shape regulatory innovation. The fund will be chaired by investor James Wise who will remain a partner at VC fund Balderton. Other staff will come from the AI Security Institute within the Department for Science, Innovation and Technology https://lnkd.in/efbY7SyX

9

2 Comments

Why Your IR35 Dispute Process Matters A clear, fair dispute process is crucial for protecting your business under the off-payroll working rules. If a contractor disagrees with your IR35 determination, you must handle the situation in a way that: ✔️Clearly explains how the decision was made ✔️Ensures the contractor has an opportunity to appeal ✔️Protects your business from future scrutiny by keeping detailed records A disputes process is mandatory, and should be transparent, repeatable, and well-documented, showing HMRC that you’ve taken reasonable care. One way to make managing disputes easier and more reliable is by working with a provider who includes dispute management as part of their process. By doing so, you benefit from their expertise in handling disputes in a fair and consistent manner, ensuring that all decisions are clearly documented. This means you’ll have a solid paper trail that can stand up to any scrutiny from HMRC. Unsure if your dispute process is set up correctly? Feel free to get in touch with us to have a chat. https://buff.ly/3IiKCcg #IR35 #IR35assessment #IR35Compliance #Compliance #Consultation #HRnews #HR #SDS #HMRC #Tax #Contractor #Contracting #Status

2

It's such a crucial step that so many founders (and more experienced entrepreneurs) regularly fail to do or fail to do in a structured way. For any early stage tech founders in my network, check out this programme, it won't disappoint.

1

Reading the NHS 10 year plan again, it's clear that the authors see patient feedback mainly as data (PROMS, PREMS, ratings etc) to be digested by various machines (AI, funding formulas), hopefully leading somehow to better care for patients. But this is to completely miss the power of narrative feedback (aka stories) to build trust and understanding, reduce defensiveness, boost morale, connect to purpose, suggest solutions and motivate change. My 20 years at Care Opinion have convinced me that narrative patient feedback isn't just information, it is an intervention which can directly benefit both patients and staff. And there's more and more research to confirm this. https://lnkd.in/ez4mw69M

305

74 Comments

If AI feels exciting but confusing then the National AI Centre has something to help! The "Guidance for AI Adoption: Foundations" published earlier this week by National AI Centre makes it easier than ever for businesses to get started with AI. Heres the secret sauce: 1️⃣ Decide who’s in charge: Pick someone to own AI in your business and write a simple policy. 2️⃣ Understand your impact: Think about who your AI might affect, what could go wrong, and how people can speak up if it does. 3️⃣ Manage risks: Spot issues early, track what happens, and fix things fast. 4️⃣ Be open: Tell people when they’re interacting with AI and keep a list of where you use it. 5️⃣ Test and check: Try it before you trust it. Keep an eye on it over time. 6️⃣ Keep humans in control: Always make sure a person can step in, pause, or switch it off if needed. Step 5 (Testing) is probably one of the hardest steps. You never know what someones going to throw into the machine and what it will spit out. Don't skimp on TESTING! If you get the steps above right (its not always as easy as it sounds), then you'll be punching above your weight. You'll build trust, accountability, and resilience as AI becomes part of daily operations. When you’re ready to go deeper, there’s a follow-up guide too! It gives step-by-step instructions for bigger or riskier projects. Innovation and transformation is still hard work, but it’s never been this accessible.

2

Had such a great time joining Craig on The Sim Tech Podcast with Ahmed 🎙️ We chatted about what we’re building at MedAscend and how AI avatars can help make communication-based simulation more accessible, realistic, and useful for learners - and even did a live demo! Thanks so much for having us, Craig! 🙌🏼

17