AutoCAD Lite

With the burgeoning success of AutoCAD, I was increasingly worried about new entrants in the market turning our own strategy against us--entering the market with a low priced package and taking the entry-level user away from us, just as we had done with the big CAD companies. The growing success of the low-priced software market, pioneered by Borland's Turbo Pascal, only intensified this concern.

This proposal, for ``AutoCAD Lite'', was the first real expression of the need for an entry level drawing product in our product line. Of course, we never did implement AutoCAD Lite, but it's interesting to compare its specifications with those of AutoSketch, which we announced to the world almost precisely two years later. In November 1993, almost a decade after this memo was written, Autodesk announced AutoCAD LT, a $495 subset of the then-current AutoCAD Release 12 (see page ).

Almost since we made our first sale of AutoCAD, we've been periodically kicking around the idea of selling a stripped version of AutoCAD at an entry-level price. These plans have always foundered as we failed to identify a usable subset of AutoCAD which could be sold in, say, the $250 to $500 range.

I believe that such a product exists, can be used for serious work, is not likely to hurt the sales of our main AutoCAD product, and can be added to our product line with minimal development effort. The product is AutoCAD-80.

Over 100 copies of AutoCAD-80 have been sold to date, and many are being used by such people as Eric Clough, who is making better use of this   limited product than most people make of the full AutoCAD-86. Thus I think that there is no question that the package is useful.

Conversely, reviewing the thin AutoCAD-80 manual, and remembering that having been cut out of the development mainstream at release 1.2, AutoCAD-80 lacks many features of even the non-ADE AutoCAD-86 that a professional is likely to require. Thus, if we assume that most of our current users are those serious professionals we talk about, few of them would spend all day working around the limitations of AutoCAD-80 to save 1750 bucks.

I am not proposing that we work with the PL/I version of the program! Instead, I am suggesting that we take the exact feature set of that program as the definition of our AutoCAD Lite product and make the C version down-configurable for that version. I would propose making some of the following strategic changes in the definition of the product. Features unavailable in AutoCAD-80 are not mentioned in the following table.

• Make the Lite version write its database out in single-precision. Greg's SCATTER/GATHER code should make this easy to do. It would halve the database size (making the package more appropriate for floppy systems), and would restrict accuracy to 8 digits or so (the same as AutoCAD-80, which is a single precision floating point package).
• No dimensioning would be available.
• Disable 8087 support.
• Remove DXF input and output capability.
• Remove LAYER capability.
• Remove LTYPE capability.
• Remove TABLET mode.
• Disable shape compiler-allow only fixed fonts.
• Develop a conversion capability to allow regular AutoCAD to read Lite drawings. Full AutoCAD drawings could not be down-converted to the Lite version.
• (I'm not sure about this) Remove pen plotter support and support only dot matrix printers.

Many of the above product definition changes are intended to turn off features which require a lot of customer support time as much as to limit the usability of the package.

The resulting product would be far more usable for the kinds of things people attempt to do with other entry-level programs like Draft-Aide,   Robocom, Caddraft, and Cascade I, but should not present serious competition in our present market. Providing an upward migration path to full AutoCAD would lock in the entry level user who purchased the Lite package. (We should, however, think seriously about the possibility that some of the manufacturers and/or Summit might buy this package to bundle as AutoCAD graphics rather than the full package. Would this be bad in the medium run, if purchasers upgraded at dealer price or list? I think some exploratory talks are needed if we decide to do this.)

Strategically, we should separate the Lite version from its big brother. We want the Lite purchaser to buy because of the association with AutoCAD, but our current customer not to seriously consider the Lite package for his purposes. Might the Lite package be well suited to schools where we have to cut very low-price deals? This might be a way to keep our price up on the main package but essentially give away unlimited Lites to cut the deal. Secondly, we might let Lite out into mass distribution and discounting. We wouldn't make much on each one, but the numbers would add up, and every Lite is a foot in the door for a full AutoCAD, and a sale we should be able to make if we follow up the Lite leads. Lite might be the foot in the door with people like Koala who don't understand the concept of $1500 software.

We have heard that the entry level packages have not been selling well, but of course one has to remember that most of them are terrible. We know that AutoCAD-80 isn't, because its users seem to be happy and productive (but wish, of course, they had all those neat ADE features...). My idea on the Lite would be to bring it out on a subset of machines (say, the ones Lotus supports), and immediately look for distribution through Softsel, Lifeboat, Koala, Computerland HQ, etc. We would package it cheaply, provide a limited number of machines, and plan no development at all after introduction so ongoing manufacturing and support costs could be low. We would not add hardware protection, but might use something like ProLok.

Development of such a product would be a moderate amount of work, as it would involve putting a lot of tests all over the product, removing a lot of code by conditional compilation and linking to make the thing small, and of course testing the final product. But once a version was certified fully debugged on a given machine the development effort would be zero. Porting to new machines would involve simply remaking the package with our regular driver for that machine. We could create the manual by editing the existing manual.

I am confident that a useful product can be created according to these guidelines. What we need to establish is:

Will it hurt our existing AutoCAD sales, and if so how much?
How much support will the package take?
Where should such a product be priced?
How much can we expect to make on each one?
What are the best distribution channels?
How much will it cost us to find out if this is a dumb idea?
Do we have the time and money to launch such a product?

Finally, there's the Japanese invasion of the low-end computer market with Z-80 machines running MSX (Microsoft's Z-80 clone of CP/M). If we entered this market with AutoCAD-80, we would then have a compatible 8086 product to fill out our product line.

So in one sentence, I think we can make such a product. Do we want to?      

AutoSketch has more than fulfilled the goals set forth in the original ``AutoCAD Lite'' proposal. As with AutoCAD, AutoSketch users immediately began to apply it in creative ways we never anticipated, and created remarkable drawings with it. This mountain bike was drawn with AutoSketch by Larry Dea.