What are the opening hours?

Monday 08:00 - 17:00 Tuesday 08:00 - 17:00 Wednesday 08:00 - 17:00 Thursday 08:00 - 17:00 Friday 08:00 - 17:00

LarcAi

LarcAI is a specialised software product & service company, a new startup in the AI field – fast m

We help organizations to open up new opportunities for innovation, productivity and better information. Robotic Process Automation (RPA) software “robots” is used to perform repetitive business processes by mimicking the way that people use their computers and make decisions. Complete business processes can be automated from start to end by software robots with limited human assistance, usually only to manage exceptions. We build Intelligent Process Automation(IPA) robots by combining Robotic Process Automation (RPA) and Artificial Intelligence powered cognitive services such as Optical Character Recognition (OCR), computer vision, language(NLP), speech, predictive analytics, tacit object models (emulate the decision-making ability of a human expert), virtual assistants and many more to drive your competitive advantage. This convergence of technologies produces automation capabilities that dramatically elevate business value and competitive advantages for customers. LarcAI partners with UIPath, a Robotic Process and intelligent Automation software vendor. UIPath powers the intelligent process automation solutions offered by most technology savvy BPO providers and Shared Service Organizations. Their mission is to eradicate tedious, redundant tasks and let software robots do the grunt work. UIPath enable businesses and companies to develop an agile robotic workforce by providing a state-of-the-art platform for software robots orchestration. We also partner with IBM on the Watson Cognitive API’s and can easily utilize Microsoft Cognitive Services APIs. Merlynn Technologies, an innovative South African company who developed Virtual Expert technology, partner with us to enable their technology used to clone human decision making. We integrated this technology with UIPath enabling robots to mimic hard human decisions.

14/10/2026

Your contact centre may be established. Your customer experience does not have to stay that way.

You have invested in phone systems, CRM and the people who run them. LarcAI adds AI agents that answer calls naturally, collect documents over WhatsApp and hand over to your team with the full story, working with the systems you already have.

No rip-and-replace required.

Talk to LarcAI about modernising your customer journey.

Contact us at :
[email protected] / 066 475 1362
[email protected] / 067 455 3288

21/09/2026

A fantastic day out at the UiPath Customer Golf Day 2026! ⛳

We were proud to be a sponsor and enjoyed connecting with customers, partners, and industry peers while sharing great conversations on and off the course.

Thank you to everyone who made the day such a success. We’re already looking forward to the next one!

10/09/2026

⛳ LarcAI is proud to be a sponsor of the UiPath Customer Golf Day 2026!

As a UiPath Platinum Partner, we're passionate about helping organisations unlock the power of automation and AI. We're looking forward to celebrating with our customers and partners at Eagle Canyon Golf Estate on 18 September 2026.

Whether we're discussing the future of automation, sharing success stories, or enjoying a day on the fairways, we're excited to connect with the incredible UiPath community.

See you on the green!

04/09/2026

As AI agents increasingly interpret corporate policy, handle customer queries, and influence decision-making, they start to look less like software and more like regulated actors within your business. 💼🤖

If an auditor walked into your office today and asked about your company's AI systems, could you confidently answer these 4 questions?

1️⃣ Can you enforce data minimisation? (Is your AI restricted to only the data it needs for that specific task?)
2️⃣ Can you demonstrate purpose limitation? (Are you preventing the AI from making secondary inferences across unrelated datasets?)
3️⃣ Can you delete data on request? (If a customer exercises their "right to erasure" under GDPR or POPIA, can you find and purge their data from your vector databases and conversation logs?)
4️⃣ Can you explain automated reasoning? (If an AI decision is challenged, can you trace why it made that choice?)

If your answer to any of these relies on manual intervention after the fact, your AI governance architecture is fragile.

In his new piece, Coenraad Reichel breaks down exactly why we need to translate legal principles (like GDPR Article 25 and POPIA Section 19) directly into technical design choices.

Don't let compliance be the bottleneck to your innovation. Build it into the foundation.

Read the full LinkedIN Article here:
https://www.linkedin.com/feed/update/urn:li:activity:7465273125929353216

Contact us at :
[email protected] / 066 475 1362
[email protected] / 067 455 3288

20/08/2026

Ask a vendor if their AI is safe and you'll get a confident yes. The better question: do they already have a document proving otherwise, sitting unread?

Two cases this year say yes. A peer-reviewed oncology study showed hidden instructions in medical scans pushing GPT-4o's lesion-miss rate from 22% to 89%. OpenAI's own GPT-5.6 system card classified its "Sol" model's file-deletion behaviour as severity level 3, two weeks before the public incidents broke.

Neither vendor was blindsided. Both had already told themselves, in writing, exactly what would go wrong.

This is the diligence question we put to every AI deployment we govern: not "could this fail," but has anyone actually read what the vendor's own testing says can go wrong, before go-live, not after the incident report.

Article 2 in our Board Blind Spots series sets out the four-field Pre-Release Warning Register we ask clients to complete before any AI system goes live.

Read the full article on our LinkedIN Page:
https://www.linkedin.com/pulse/openais-unheeded-warnings-how-pre-release-data-foretold-recent-q4nff

Contact us at :
[email protected] / 066 475 1362
[email protected] / 067 455 3288

19/08/2026

Is document intelligence an extraction problem or a pipeline design problem?

If you’re perpetually chasing a better accuracy metric but still seeing bottlenecks, it’s likely the latter. Scaling to hundreds of documents a day requires more than just a good OCR engine; it requires an architecture that is:

✅ Scalable
✅ Adaptive
✅ Resilient

As we wrapped up in our recent feature, the number (99%) isn't the achievement … the architecture that makes it reliable is.

Contact us at :
[email protected] / 066 475 1362
[email protected] / 067 455 3288

18/08/2026

"95% of AI projects fail."
"85% of AI projects fail."

You've probably seen both statistics on strategy slides, board packs, and executive presentations.

But here's the uncomfortable question:

Who actually counted?

In researching this piece, I traced two of the most quoted AI statistics back to their original sources. One was transformed from a narrow research finding into a universal truth. The other became a boardroom fact despite the original source never making the claim.

The issue isn't whether AI succeeds or fails.

The issue is that boards are making capital allocation, risk, and strategy decisions based on numbers that are often:

✅ Unverified
✅ Misquoted
✅ Detached from their original context

We would never accept that standard for financial data.

Why do we accept it for AI?

In this first article of the Board Blind Spots series, I introduce the Stat Provenance Test: a simple framework boards can use to assess whether an AI statistic deserves a seat at the decision-making table.

Before the next AI number influences a major investment, ask:

Who counted, how, and may we see it?

Read the full article on our LinkedIN Page:
https://www.linkedin.com/pulse/provenance-gap-auditing-ai-statistics-your-board-pack-larcai-ckbkf

Contact us at :
[email protected] / 066 475 1362
[email protected] / 067 455 3288

04/08/2026

95% of enterprise GenAI pilots show no measurable financial return. That statistic has stopped being controversial and started being wallpaper.

The interesting question is what the working few did differently. It was not a better model. They asked a narrower question, fixed the data before the demo, and wrote the baseline down before they believed anything.

Article 4, the finale of our For the Record series, walks through it and ends with a one-page ROI Baseline Memo: the record you sign before build that turns "is this working" into an arithmetic problem instead of a debate.

LinkedIN Article:
https://www.linkedin.com/pulse/what-5-did-larcai-c8ppf

Contact us at :
[email protected] / 066 475 1362
[email protected] / 067 455 3288

03/08/2026

The sandbox held. The agent still got owned.

Recent security research showed AI coding agents hijacked through the content they read, a repository, an error event, a skill file, and not through any breached permission. Meanwhile large enterprises are heading for more than 1,600 agents each, and fewer than one in five keep an inventory.

You cannot revoke an identity you never issued, or investigate a log that blames one shared account.

Article 3 in For the Record extends our Blast Radius Ladder with the new evidence, and ends with a one-page AI Agent Inventory that answers the question every auditor asks first: how many agents are you running?

LinkedIN Article link:
https://www.linkedin.com/pulse/agent-zero-trust-1600-agent-problem-larcai-5whff

Contact us at :
[email protected] / 066 475 1362
[email protected] / 067 455 3288

30/07/2026

The sandbox held. The agent still got owned.

Recent security research showed AI coding agents hijacked through the content they read, a repository, an error event, a skill file, and not through any breached permission. Meanwhile large enterprises are heading for more than 1,600 agents each, and fewer than one in five keep an inventory.

You cannot revoke an identity you never issued, or investigate a log that blames one shared account.

Article 3 in For the Record extends our Blast Radius Ladder with the new evidence, and ends with a one-page AI Agent Inventory that answers the question every auditor asks first: how many agents are you running?

LinkedIN Article Link:
https://www.linkedin.com/feed/update/urn:li:activity:7487763787878486017

Contact us at :
[email protected] / 066 475 1362
[email protected] / 067 455 3288

Address

Soetdoring Building, 7 Protea Street
Doringkloof
0157

Opening Hours

Monday	08:00 - 17:00
Tuesday	08:00 - 17:00
Wednesday	08:00 - 17:00
Thursday	08:00 - 17:00
Friday	08:00 - 17:00

Website

http://larcai.com/

Alerts

Be the first to know and let us send you an email when LarcAi posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.

Contact The Business

Send a message to LarcAi: